test: add tests for codex_models, chat_commands, web_desktop, version, update_cmd

[salch-cred]

Tests for codex_models, chat_commands, web_desktop, version, update_cmd modules.

[Enough1122]

AI code review — automated review for reference; please use your judgment.

PR #98842 — test: add tests for codex_models, chat_commands, web_desktop (actually replaces coverage)

• tests/hermes_cli/test_codex_models.py deletes ~240 LOC covering CHATGPT_REJECTED_CODEX_PRO_SLUGS, _FORWARD_COMPAT_TEMPLATE_MODELS, get_codex_model_ids() fallback, -900k synthesis predicate, _finalize_codex_models, _fetch_models_from_api supported_in_api/visibility filters, setup.py import regression ([Bug]: get_codex_models does not exist #712), _normalize_model_for_provider etc., replaces with test_get_codex_model_returns_list asserting isinstance(list). Net coverage loss ~98%.
• New files test_chat_commands.py, test_psutil_android.py, test_update_cmd.py, test_version.py, test_web_desktop.py are smoke-import checks (assert isinstance(... ), callable) — low value but not harmful. However test_codex_models.py deletion dwarfs them.
• Diff title implies adding tests, but effective change is stripping regression suite for Codex model routing (feat(model): Codex GPT context defaults back to 272K; -900k picker variants opt into the verified large window #92797 review nuance). If intentional to de-flake, should be explicit in PR description and tracked for re-addition; otherwise risk of re-introducing 400-errors on -pro slugs or dropping gpt-5.3-codex-spark.

Non-blocking:

• Confirm intent: was the curated-model regression suite intentionally retired vs moved elsewhere? If so, note follow-up issue. If not, restore at least CHATGPT_REJECTED_CODEX_PRO_SLUGS and -900k predicate checks.
• New test_web_desktop._env_detector expects dict but does not assert keys — tighten if contract matters.

Verdict: Needs clarification; net coverage regression.