feat: add Atlas Cloud LLM provider

[binyangzhu000-sudo]

Summary

• add Atlas Cloud to the backend LLM provider registry as an OpenAI-compatible provider
• support atlas and atlas-cloud aliases with the Atlas Cloud default base URL
• add Atlas Cloud to the enterprise settings fallback provider list
• add focused provider registry, factory, token default, and payload-shape tests

Checklist

• Tested locally
• No unrelated changes included

Validation:

• uv run --python 3.11 --with pytest --with pytest-asyncio --with httpx --with loguru --with pydantic --with pydantic-settings --with sqlalchemy --with asyncpg --with python-jose --with passlib --with python-multipart --with PyYAML --with redis --with croniter --with aiofiles --with docker --with websockets --with pypinyin --with anyascii --with pillow --with beautifulsoup4 --project backend pytest tests/test_atlas_cloud_llm_provider.py tests/test_llm_system_message_shape.py (19 passed)
• uv run --python 3.11 --project backend python -m compileall app/services/llm/client.py tests/test_atlas_cloud_llm_provider.py
• npm ci --ignore-scripts
• npm run build
• npm test (85 passed)
• git diff --check && git diff --cached --check

Notes:

• No README, docs, logo, sponsor, credits, or partner content was changed.
• The Atlas Cloud model catalog endpoints returned HTTP 403 with the current local key during validation, so the PR only uses the documented OpenAI-compatible base URL and model IDs.