feat(cli): /hooks preview <event> lists hooks that would fire

[emal-avala]

Summary

A misspelled event in `settings.toml` (e.g. `pre-tool-used` instead of `pre_tool_use`) silently deserializes as a hook that never matches anything at runtime. Users only notice when the hook seems not to run — often hours of debugging later. And even with correctly-spelled events, there's no way to audit which of my configured hooks would actually fire for `user_prompt_submit` without tracing the code.

This PR adds a `/hooks preview ` subcommand:

```

/hooks preview pre_tool_use
Hooks that would fire for 'pre_tool_use' (2):

1. shell: ./check-bash.sh (tool=Bash)
2. http: POST https://my-audit.internal/event
   ```

• Shows "No configured hooks match event ''" when nothing matches.
• Rejects unknown event names with a hint pointing at `/hooks events`.
• Accepts both canonical `snake_case` and the hyphen / mixed-case forms users tend to type (`pre-tool-use`, `Pre-Tool-Use`).

Also adds a round-trip test that pins the new `parse_hook_event` helper to `HOOK_EVENT_CATALOG` — adding a new event to the catalog without teaching the parser will now fail the build instead of silently rejecting that event name from preview.

Test plan

• `cargo clippy -p agent-code --tests --no-deps -- -D warnings` — clean
• `cargo test -p agent-code --bin agent commands::tests::` — 85 pass (80 prior + 5 new)
• 5 new tests: canonical snake_case parsing, hyphen/case tolerance, whitespace trimming, unknown-name rejection, catalog round-trip

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.