fix: handle terminal startup failures and degraded session watchers

[danshapiro]

Summary

• surface provider-aware startup and restore failures instead of reconnect loops that flash and reload dead tabs
• return exited-terminal attach errors with recent output so restores fail in place with useful context
• fall back to periodic coding-session full scans when chokidar watch setup degrades or late roots are still missing

Testing

• npm run test:vitest -- test/unit/client/components/TerminalView.lifecycle.test.tsx
• npm run test:vitest -- --config vitest.server.config.ts test/server/ws-edge-cases.test.ts test/unit/server/terminal-registry.test.ts test/unit/server/coding-cli/session-indexer.test.ts
• npm test

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c0ef18bbe2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Avoid debouncing away every fallback full scan

The fallback loop schedules a refresh every rootFallbackScanIntervalMs using scheduleRefresh(), but scheduleRefresh() always clears any pending timer before creating a new one. With the defaults in this commit (rootFallbackScanIntervalMs = 1000, debounceMs = 2000), each fallback tick cancels the prior refresh before it can fire, so no full scan runs while fallback mode is active. In degraded watcher or missing-root scenarios, this prevents the intended recovery behavior and sessions may not appear until some unrelated trigger/full-scan interval occurs.

Useful? React with 👍 / 👎.