Redesign Pushgate CLI output

[dbrosio3]

This pull request introduces several improvements to both the user experience and internal logic of the pushgate tool, focusing on enhanced output formatting, better post-push guidance, and improved runner handling. The changes include a new post-push summary with pull request links, more readable and structured AI review output, and updates to the pre-push hook for better runner invocation and diagnostics.

User interface and output improvements:

• Added a post-push summary after successful branch pushes, including upstream info and a direct GitHub pull request link when applicable. This involves new logic for parsing push arguments, resolving branch and remote details, and formatting the summary output (src/git/push.ts, src/cli.ts). [1] [2] [3] [4] [5]
• Refactored local AI review transcript output to use structured, color-coded, and more readable sections and result rows, replacing plain text log lines. This includes new use of formatting helpers and clearer messaging for skipped, blocked, warning, and passed states (src/ai/transcript.ts). [1] [2]

Pre-push hook and runner handling:

• Updated the pre-push hook to support both Node.js and native runners by detecting the runner type and invoking with node if necessary. Also, added a PUSHGATE_VERBOSE environment variable to control diagnostic output, making routine hook runs quieter and verbose mode available for troubleshooting (hook/pre-push, README.md). [1] [2] [3] [4] [5] [6] [7]

Error handling and diagnostics:

• Improved error messages for failed git diff operations in the AI review context, including exit code and signal for easier debugging (src/ai/review-context.ts).

These changes collectively improve the clarity, usability, and maintainability of the tool, while also making it easier for users to understand push results and AI review outcomes.