docs: record motion plan deviations

Co-Authored-By: Claude Opus 5.5 <[email protected]>

docs/superpowers/plans/2026-10-30-v11-motion.md

@@ -207,3 +207,11 @@
 - [ ] Run it against the dev server on port 3491 (with `STC_DATA_DIR=/tmp/stc-v11-motion`). Look at every frame.
 - [ ] Run `bun test`, `bun run typecheck`, and the PG run. Check that `bun.lock` and the `package.json` dependencies are unchanged.
 - [ ] Write the report to `/home/user/STCourses/.superpowers/sdd/2026-10-30-stcourses/v11-motion-report.md`. Commit `chore: motion frame capture script`.
+
+## Deviations found during execution
+
+- **`flushSync` on route changes became `v7_startTransition`.** With `flushSync`, the first visit to a lazily loaded route captured an empty page, so the sidebar dropped out and popped back in. Router updates now run in a React transition: the current page stays up until the new route has rendered, then the view transition captures it.
+- **`::view-transition { pointer-events: none }` has no effect.** Chrome sends pointer input to the document element while a transition runs. Instead, each transition is kept to the old page's 120 ms fade, and the root group's default 250 ms morph is turned off. Keyboard input is unaffected.
+- **`*` does not match `::details-content`.** The reduced-motion block has a separate `*::details-content` rule.
+- **Chrome lists no `::details-content` animations in `getAnimations()` and sends no transition events for them.** The fold tests sample the computed height instead.
+- **Pre-existing bug: pdf.js 6 documents have no `destroy()`.** Leaving a lesson whose deck had loaded crashed the app. Fixed by destroying the loading task instead.