feat(config): add named configuration profiles

[emal-avala]

Summary

Adds named config profiles. `/profile save|load|list|delete` persists `Config` snapshots under `<config_dir>/agent-code/profiles/.toml`. Loading a profile replaces the runtime config wholesale.

Commands:
```
/profile list saved profiles
/profile list (same)
/profile save save current config
/profile load replace runtime config
/profile delete remove
/profile help usage
```

Design notes

• Full replacement, not merge. Profiles are snapshots. Merging two user-maintained configs is where most "why is this setting ignored?" bugs come from — this keeps the mental model trivial.
• Env vars on load. `AGENT_CODE_MODEL` / `AGENT_CODE_API_BASE_URL` / API keys from env are NOT re-applied when a profile is loaded (the process keeps using whatever env said at startup; the profile's TOML-declared values override the file, not the env). Documented in the load output so it's not surprising.
• Name validation is load-bearing. `validate_name` rejects empty, oversize (>64 chars), path separators, `..`, and shell metachars. A malicious or careless name cannot escape the profiles directory.

Files

• New `crates/lib/src/services/profiles.rs` with `save_profile`, `load_profile`, `list_profiles`, `delete_profile`, `ProfileSummary` (~180 LOC)
• `crates/lib/src/services/mod.rs` — new `pub mod profiles`
• `crates/cli/src/commands/mod.rs` — new COMMANDS entry + dispatch + `execute_profile` helper

Test plan

• `cargo fmt --all` clean
• `cargo check` passes
• `cargo clippy --no-deps` clean
• `cargo test -p agent-code-lib --lib profiles` — 5 tests on name validation (accepts safe names, rejects empty / path escape / special chars / oversize)
• Manual: `/profile save work` → file at `/agent-code/profiles/work.toml`
• Manual: `/profile list` shows `work` with its model
• Manual: `/profile load work` swaps the in-memory config
• Manual: `/profile save ../etc/passwd` → rejected with a clear error, no file written outside profiles dir

[chatgpt-codex-connector]

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