feat: support GPT-6 Astra, GPT-6 Sol, and GPT-6 Luna

[fabriciomalta]

Summary

• Request the newer Codex model catalog and ignore caches from older catalog versions.
• Add GPT-6 Astra, Sol, and Luna to static model registration and priority tier fallback.
• Update the README model list and cover OpenAI/Ollama listing plus cache invalidation.

How to try locally

1. Run python -m unittest discover -s tests -q.
2. Start ChatMock with model sync enabled and request GET /v1/models.
3. Confirm gpt-6-astra, gpt-6-sol, and gpt-6-luna appear when available to the signed-in account, then send a short Chat Completions request for each.

The local test suite passed (19 tests). No routing, CLI, GUI, Docker, or public parameter changes are needed.

Closes #125

[Copilot]

Copilot review overview

🟢 Approval recommended

The only outstanding feedback is a minor fast-mode test-coverage nit with no blocking issue identified.

Review effort: Lite
Findings: None

What changed in this PR

Adds GPT-6 Astra, Sol, and Luna support across model discovery, registration, fallback behavior, documentation, and tests.

Changes:

• Updates catalog versioning and stale-cache invalidation.
• Registers GPT-6 models with reasoning and priority support.
• Expands model-listing and cache tests and updates the README.

File	Summary
tests/​test_routes.py	Covers OpenAI and Ollama model listings.
tests/​test_model_catalog.py	Covers catalog cache invalidation.
README.md	Documents the supported GPT-6 models.
chatmock/​model_registry.py	Adds static GPT-6 model registrations.
chatmock/​model_catalog.py	Updates catalog version and cache validation.
chatmock/​fast_mode.py	Adds priority fallback support; a minor request remains for direct fast-mode assertions.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[EmBista]

Thanks for putting #126 together. I think adding the GPT-6 IDs fixes the immediate symptom, but it leaves us needing another code change whenever the model list moves.
I’ve opened #127 as an alternative: it takes the client version from the latest stable Codex release and builds the advertised model list from the account catalogue.
That should let new models appear without adding them to a hard-coded list. I’d welcome your thoughts on that approach.