refactor: centralize cleanup and risk threshold

[Hiten0305l]

Summary

This refactor removes duplicated logic by centralizing shared functionality used by both the CLI and the MCP server. It ensures that the risk-score threshold and temporary directory cleanup logic are defined in a single place, making future maintenance easier and preventing the implementations from drifting.

Changes

• Moved the shared RISK_THRESHOLD constant to skillspector.constants.
• Updated both cli.py and mcp_server.py to import and use the shared RISK_THRESHOLD instead of maintaining duplicate values.
• Added a shared cleanup_result() helper in skillspector.cleanup to centralize temporary directory cleanup.
• Replaced the duplicated cleanup logic in both cli.py and mcp_server.py with the shared helper.
• Removed duplicated implementations while preserving the existing CLI and MCP server behavior.

This change is purely a refactor and does not modify existing functionality.

Validation

• Verified that RISK_THRESHOLD is defined only once and imported where needed.
• Verified that temporary directory cleanup is centralized and used by both the CLI and MCP server.
• Verified that the optional MCP dependency installs successfully using uv sync --extra mcp.
• Existing functionality remains unchanged.

Fixes #192

[rng1995]

This refactor preserves both threshold semantics (CLI fails above 50; MCP treats 50 and below as safe) while removing duplicate cleanup logic. The shared helper matches the prior behavior and is used by the scan, baseline, and MCP paths. I found no correctness or security blocker.

[Hiten0305l]

Thanks for the review and approval! I noticed there's still a workflow awaiting maintainer approval. Please let me know if there's anything else needed from my side. Thanks!