feat(native-chat): add Muse live-terminal effort picker

[clichedmoog]

Stacked on #22998 — review only the top commit (feat(native-chat): add Muse live-terminal effort picker) until #22998 merges, then this rebases onto main.

ELI5

Muse chat sessions grow an effort pill: pick Minimal–Ultra in the composer and the live terminal session switches reasoning effort via /effort, the same way Claude's /model switch works.

What Changed

• MUSE_EFFORT gains midSession: { kind: 'command', build: (value) => '/effort <value>' }, reusing the 7-level vocabulary feat(orchestration): accept Muse model and effort for supervised workers #22383 defined for worker launches. Same shape as the antigravity and claude-codex effort options.
• New readMuseSessionOptionsFromTerminalScreen: scrapes the TUI status line (<model> · <effort> · <cwd> · <mode>, bottom-up after ANSI strip) and reports { model, effort } so the pill shows what the session actually runs. The model id is reported raw — ids are account-scoped with no listing command, so there is still no model picker by design.
• use-native-chat-session-options reports those values for muse terminal tabs, mirroring the Claude hookup.

Why

#22998 renders the conversation but leaves the composer pill-less; effort is the one Muse session option with a fixed, switchable vocabulary. Alternatives: worker-launch-only (already exists, does not help live sessions); a model picker (impossible — no listing command); polling session.jsonl for effort (the log has no effort field; the status line does).

Linked Issue

Fixes #23088

Visual Proof

Desktop dev build on macOS, live Muse session whose TUI status line reads muse-spark-1.3-contributor · max · …:

TUI status line (source)	Chat composer (reported pills)

Picking from the pill dispatches into the live PTY and the pill tracks the result:

This also confirms the /effort <level> slash syntax against Muse 1.4.0.

Testing

• I manually tested these changes locally (dev build, macOS: pill reports live effort, picks dispatch /effort <level> into the PTY)
• Automated tests added/updated

Reader (5 tests) and catalog dispatch (3 tests). pnpm tc, oxlint/oxfmt on touched files, and the changed-lines quality gate pass locally.

AI Disclosure

Written with AI assistance (Muse Code, Meta Muse Spark); verified with the repo's own test suites.

Agent skill upstream boundary

• Not applicable — no skill installer source, tests, fixtures, registry entries, path tables, comments, or documentation touched.

Checklist

• This PR is small and focused
• I explained what changed and why (ELI5, the user-facing before/after, the mechanism, and why over the alternatives)
• Before/after screenshots or videos attached for UI changes, or N/A with reason
• Self-reviewed for correctness, security, and performance
• Cross-platform, SSH/remote, and path/shortcut impact considered (pure string parse of the shared status-line format; PTY command delivery identical to existing switches)
• pnpm lint, pnpm typecheck, pnpm test, and pnpm build pass (or CI will cover; local preferred)