Investigate local code review tooling to reduce PR review iterations #35

Summary

Investigate options for running automated code review locally before pushing, to reduce the number of PR review iterations.

Background

PR #33 went through 8 rounds of Copilot review with 28 total comments. Many issues (edge cases, naming, platform concerns) could have been caught before the first push if a local review step existed.

Options to investigate

1. gh copilot CLI — The general-purpose Copilot CLI (gh copilot) exists but has no review subcommand. Investigate if it can be prompted to review a diff or staged files for code quality.

2. GitHub Copilot code review API — Check if Copilot pull request reviewer can be invoked programmatically on a local diff before creating a PR.

3. Custom pre-push script — A script that runs the diff through a local LLM or the Copilot API with the repo's coding guidelines, then blocks the push if issues are found.

4. Roslyn analyzers — We already have StyleCop and Roslynator. Consider adding:

   • Custom analyzers for single-letter lambda detection
   • Custom analyzers for ConfigureAwait(false) without named parameter
   • These would catch issues at build time, not just in pre-commit grep

Acceptance criteria

• Evaluate at least 2 options above
• Implement the most practical option
• Demonstrate it catches at least 3 of the issue types from PR Resolve SDK from workspace project NuGet references (no nupkg required) #33