Record silent WebM screen video instead of GIF during recordings ...

The recorder now captures continuous screencast frames via the CDP
screencast API and encodes them into a silent WebM (VP8, ~15 fps, capped
at 800px wide) uploaded as recording.webm. WebCodecs VideoEncoder is not
exposed to service workers, so encoding runs in an offscreen document
(chrome.offscreen) fed by base64 frame batches over a runtime port; the
result is verified with a local <video> playback probe and falls back to
the cursor-following GIF when encoding or verification fails (Firefox,
old Chrome, broken output).

Root cause of the earlier corrupt output: webm-muxer silently produces a
broken file when the first chunk's timestamp is non-zero — media
timestamps now start at the first frame and the muxer uses
firstTimestampBehavior: "offset".

The e2e now validates the uploaded recording with ffprobe (vp8 codec,
positive duration) and still covers the GIF fallback path, console-step
dump, navigation survival and replay.

Co-Authored-By: Claude Code <noreply@anthropic.com>

Extension: popup UI, panel URL, icon fix, GIF tracks and track replay ...

- Popup replaces the auto-start picker: mode choice (element note /
  record steps), latest report card with open / replay / delete, web
  panel link; share links open at panelUrl (default :5173)
- Register the Phosphor font via the FontFace API — constructable
  stylesheets ignore @font-face, which left overlay icons empty
- Recorder tracks upload as one animated GIF (gifenc) instead of per-
  step screenshots; frames follow event pacing, UI hidden on capture
- Replay a recorded track from the popup: background opens the start
  page, re-executes click/input/url_change steps with original pacing
  and flashes the result in the overlay

Co-Authored-By: Claude Code <noreply@anthropic.com>

Extension: picker, note composer, recorder + infra (Makefile, prod compose, README) ...

- MV3 extension for Chrome and Firefox: single codebase, per-entry IIFE
  builds (background / content / options), manifests generated per target
- Background service worker owns all networking (Bearer token from
  storage.local), captureVisibleTab (windowId!), recorder event buffer
  with offset_ms and auto-screenshots on clicks
- Content overlay in a closed shadow root (constructable stylesheets,
  kit CSS fetched from assets/style.css with /assets URL rewrite);
  PickerLayer, NoteComposer (autofocus, Ctrl+Enter submit),
  RecorderBar (step counter, note, stop)
- Options page: server URL, login, default project
- Recorder steps link screenshots by uploaded file id (server accepts
  attachment id or file id); environment/element JSONB serialized with
  mode="json" (datetime captured_at broke inserts)
- e2e harness: headless Chromium + extension, note + recording flows
- Makefile, docker-compose.prod.yml (caddy :8081 + static panel),
  deploy/Caddyfile, README (ru) with extension install and /assets notes

Co-Authored-By: Claude Code <noreply@anthropic.com>