fix(jan): map runtime capabilities - #221

Summary

Adds a pure Jan capability-normalization helper and focused unit coverage for issue #27.

Changes

• Map Jan capability values completion, chat, embeddings, vision, reasoning, and tools to the existing ModelDock Capability enum.
• Normalize case and surrounding whitespace.
• Preserve first-seen order and deduplicate values.
• Ignore missing, malformed, non-string, and unknown values without raising.
• Keep the change adapter-local; no domain, core, ports, CLI, public API, or dependency changes.
• Leave Jan model discovery/filtering to the separate Jan AI: implement runtime adapter #25/Jan AI: read local model directory #26 work because this issue does not yet have a metadata source.

Testing

• uv run --extra dev pytest tests/unit/test_jan_runtime.py -q — 13 passed
• uv run --extra dev ruff check src tests — passed
• uv run --extra dev ruff format --check src tests — passed
• uv run --extra dev mypy src — passed for 79 source files
• git diff --check — passed
• Full suite independently recorded as 551 passed; three local offline CLI fixture failures were unrelated to this change.

Checklist

• Branch named per Git Workflow (fix/jan-capability-mapping)
• Not developed on main
• Code follows AGENT.md coding standards
• domain/ and ports/ stay pure
• Quality gates pass locally: Ruff, mypy, targeted pytest, and diff check
• No docs or version changes were needed
• No secrets or generated artifacts included

Closes #27.