Add Atlas Cloud LLM provider

[binyangzhu000-sudo]

Summary

• Add Atlas Cloud as a built-in OpenAI-compatible LLM provider.
• Wire Atlas Cloud through backend config, API key handling, model discovery, translation/refine flows, and the web provider selector.
• Add focused unit coverage for provider creation, API-key mapping, and checkpoint secret handling.

Tests

• .venv/bin/python -m pytest tests/unit/test_atlascloud_provider.py tests/unit/test_api_keys.py tests/unit/test_checkpoint_secrets.py -q
• .venv/bin/python -m compileall src
• node --check src/web/static/js/providers/provider-manager.js
• node --check src/web/static/js/utils/api-key-utils.js
• node --check src/web/static/js/core/settings-manager.js
• node --check src/web/static/js/ui/form-manager.js
• node --check src/web/static/js/translation/batch-controller.js
• node --check src/web/static/js/providers/provider-select-helpers.js
• git diff --check

Atlas Cloud model catalog was also checked live: /v1/models returned 118 models and includes qwen/qwen3.5-flash plus deepseek-ai/deepseek-v4-pro.

No root README changes; no sponsor, logo, credits, or partner promotion added.