fix(highway): a SUPERSEDED renderer init is not a FAILED one (#970)

Starting a gig dropped the player onto the fallback 2D highway with no venue.

startGig() calls setViz('venue'), which installs the 3D renderer — whose init is
async — and then immediately starts its play queue. playSong() re-initialises
that same renderer a tick later. A renderer mints a fresh readyPromise per
init() and rejects the previous one with "superseded"; highway.js only checked
that the RENDERER OBJECT was unchanged, which it is. So it treated a healthy,
re-initialising renderer as a failed one, tore it down, and reverted to 2D:

    renderer async init failure: Error: superseded
    viz picker: reverted to default renderer (async-init-failure)

The guard now also checks the PROMISE identity: a rejection from an init cycle
the renderer has already moved on from is ignored. The renderer-identity guard
stays (a rejection for a renderer since REPLACED is also not ours), and a
genuine failure of the CURRENT cycle still reverts — both init() call sites go
through _setRenderer, which re-wires the handler every time, so the new cycle is
always watched.

Reproduced and fixed against the real build:

    before:  vizSelection=default  viz-picker=default  venue=inactive  viz:reverted
    after:   vizSelection=venue    viz-picker=venue    venue=ACTIVE    (no revert)

Also widens the paused-frame throttle's opt-out. The throttle fires whenever the
CHART CLOCK is stalled — not only on a pause, but through a count-in and the
credits/author overlay too. Its opt-out only asked "is a crowd video rolling",
but the venue scene animates on a clock of its own with no pack at all (backdrop
breathe, parallax, haze drift, warmth pulse — Math.sin(t) in the draw loop), so
that motion was still being throttled. It now claims frames for both sources; a
plain 3D highway with no venue reads motion mode 'off' and keeps the #654 GPU
saving.

HONEST CAVEAT on that second part: I could not get the throttle to fire in a
reproduction. A control run on the shipped code showed 100 draws/sec while
paused, not the ~10/sec a firing throttle would give — so the change is
defensible on its own terms (a stalled clock is genuinely not a static picture)
but it does NOT have a demonstrated symptom behind it. The viz fix above does.

Tests: the superseded guard, and that the throttle opt-out covers both motion
sources. All fail against the pre-fix source. eslint 0 errors; JS 1207/1207.

plugins/highway_3d/screen.js

@@ -15398,15 +15398,29 @@
             // highway throttled the whole room. Pausing the song dropped the
             // venue, the crowd and the stage to 10 fps.
             //
-            // Only claim continuous frames while a crowd video is actually
-            // rolling: with no venue pack (the common case) the paused scene IS
-            // static and the throttle should still save the GPU.
+            // Two independent sources of motion, and BOTH must keep their frames:
+            //
+            //  • a crowd video rolling on its own clock (career venue pack), and
+            //  • the venue scene's own fake-depth motion — the backdrop breathes,
+            //    the haze drifts, warmth pulses, the shimmer moves. That is
+            //    Math.sin(t) in the draw loop (see _venueApplyFakeDepthMotion),
+            //    so it only moves while we are actually given frames, and it runs
+            //    with NO pack at all.
+            //
+            // The throttle fires whenever the CHART CLOCK is stalled — which is
+            // not just a pause. A count-in and the credits/author overlay stall it
+            // exactly the same way, so the venue was stuttering there too.
+            //
+            // With no venue at all (plain 3D highway) the paused scene really is a
+            // still picture: motion mode reads 'off', we claim nothing, and the
+            // throttle still saves the GPU as #654 intended.
             needsContinuousFrames() {
                 if (!_isReady || _ctxLost) return false;
                 for (const v of _venueCrowdVideos) {
                     if (v && !v.paused && !v.ended && v.readyState >= 2) return true;
                 }
-                return false;
+                // 'off' also covers prefers-reduced-motion and "no venue scene".
+                try { return _venueEffectiveMotionMode() !== 'off'; } catch (_) { return false; }
             },
 
             draw(bundle) {

static/highway.js

@@ -986,6 +986,22 @@ function createHighway() {
                 // inline arrow function.
                 function _handleAsyncInitFailure(e) {
                     if (hwState._renderer !== _installedRenderer) return;
+                    // ...and ignore a rejection from a SUPERSEDED init cycle.
+                    //
+                    // A renderer mints a fresh readyPromise on every init(), and
+                    // rejects the previous one ("superseded") when a newer init
+                    // starts. The renderer object is unchanged, so the identity
+                    // check above does not catch it — and we would tear down a
+                    // perfectly healthy renderer that is merely re-initialising.
+                    //
+                    // This is exactly what starting a gig did: setViz('venue')
+                    // installed the 3D renderer, then the queue's playSong()
+                    // re-initialised it a tick later; init #1's promise rejected,
+                    // and the gig dropped to the fallback 2D highway with the
+                    // venue gone. A superseded init is not a failed init — the
+                    // NEW cycle owns the outcome, and its own promise is what we
+                    // must judge.
+                    if (_installedRenderer.readyPromise !== rp) return;
                     console.error('renderer async init failure:', e);
                     _destroyCurrentIfInited();
                     hwState._renderer = _defaultRenderer;

tests/js/highway_pause_throttle.test.js

@@ -114,13 +114,52 @@ test('the capability probe fails closed (absent / non-function / throwing)', ()
         'only an explicit true opts out — a truthy accident must not disable the throttle');
 });
 
-test('3D highway claims continuous frames only while a crowd video is rolling', () => {
+test('3D highway claims continuous frames for BOTH sources of venue motion', () => {
     const h3d = fs.readFileSync(
         path.join(__dirname, '..', '..', 'plugins', 'highway_3d', 'screen.js'), 'utf8');
     const fn = extractBlock(h3d, 'needsContinuousFrames()');
+    // (1) a crowd video rolling on its own clock (career venue pack)
     assert.match(fn, /_venueCrowdVideos/, 'must key off the actual crowd video elements');
-    assert.match(fn, /\.paused/, 'a paused video is a still frame — throttle should still apply');
-    // With no venue pack (the common case) the paused scene really is static and
-    // the GPU saving must survive: the method has to be able to return false.
-    assert.match(fn, /return false;/, 'must fall through to false with no live video');
+    assert.match(fn, /\.paused/, 'a paused video is a still frame');
+    // (2) the venue scene's OWN fake-depth motion — backdrop breathe, haze drift,
+    // warmth pulse, shimmer. Math.sin(t) in the draw loop, so it only moves while
+    // we get frames, and it runs with NO pack at all. Missing this meant the venue
+    // still stuttered on pause / count-in / credits whenever no video was rolling.
+    assert.match(fn, /_venueEffectiveMotionMode\s*\(\s*\)\s*!==\s*'off'/,
+        'the venue scene animates without any video — it must claim frames too');
+    // ...and with no venue at all the paused scene IS static: the #654 GPU saving
+    // must survive, so the method has to be able to return false.
+    assert.match(fn, /return false;/, 'must fall through to false on a plain 3D highway');
+});
+
+// ── a SUPERSEDED init is not a FAILED init ──────────────────────────────────
+//
+// Starting a gig dropped the player onto the fallback 2D highway with no venue.
+//
+// setViz('venue') installs the 3D renderer, whose init is async; the gig then
+// immediately starts its play queue, and playSong() re-initialises that same
+// renderer a tick later. A renderer mints a fresh readyPromise per init() and
+// rejects the previous one with "superseded" — but highway.js only checked that
+// the RENDERER object was unchanged, which it is. So it treated a healthy
+// re-initialising renderer as a failed one, tore it down, and reverted to 2D:
+//
+//     renderer async init failure: Error: superseded
+//     viz picker: reverted to default renderer (async-init-failure)
+//
+// Reproduced and fixed against the real build (venue stays selected, scene
+// active, no viz:reverted).
+
+test('a superseded readyPromise must not revert the viz to 2D', () => {
+    const src = highwaySources();
+    const fn = extractBlock(src, 'function _handleAsyncInitFailure(e)');
+    assert.match(fn, /readyPromise\s*!==\s*rp[\s\S]{0,40}return/,
+        'a rejection from a STALE readyPromise (the renderer has since re-init\'d) must be ' +
+        'ignored — otherwise a re-initialising renderer is torn down as if it had failed');
+    // The renderer-identity check must survive too: a rejection belonging to a
+    // renderer that has since been REPLACED is also not our problem.
+    assert.match(fn, /hwState\._renderer\s*!==\s*_installedRenderer[\s\S]{0,20}return/,
+        'the renderer-identity guard must remain');
+    // ...and a genuine failure of the CURRENT init cycle must still revert.
+    assert.match(fn, /_emitVizReverted\s*\(\s*'async-init-failure'\s*\)/,
+        'a real async-init failure must still fall back to the default renderer');
 });