Add MiniMax provider coverage

[octo-patch]

Reason: add target provider/model to existing provider registry

Summary

• Register MiniMax-M3 and MiniMax-M2.7 pricing from the target text model configuration, including cache rates.
• Keep cumulative cost estimates accurate per request and support LiteLLM-prefixed model names.
• Document global and China OpenAI- and Anthropic-compatible endpoints in both provider guides.
• Merge the current main branch so the PR is conflict-free.

Checks

• uv run --extra dev pytest tests/ -q (92 passed)
• uv run python -m compileall -q corecoder tests (passed)
• uv run --extra dev ruff check corecoder/llm.py tests/test_core.py tests/test_demo.py --select E,F (passed)
• Full repository Ruff still reports pre-existing baseline findings outside this change.
• GitHub Actions CI is awaiting maintainer approval (action_required); no job logs were produced.