mcp: add tool-level logging to GitHub MCP servers to expose workflow bottlenecks

## Problem

GitHub MCP server tool calls are not being logged, creating visibility gaps in workflow optimization. During `/close-issue 848` session:

- **Git MCP server calls**: All logged with timestamps in `~/mcp-tool-calls.log`
- **GitHub MCP server calls**: No logging - 1m 42s gap between `git_push` and `git_worktree_remove`

This prevents identifying actual bottlenecks in the `/close-issue` workflow.

## Evidence from MCP Tool Logs

```
2025-07-15 11:12:21 - git_push: SUCCESS
[1m 42s gap - GitHub API calls invisible]
2025-07-15 11:14:03 - git_worktree_remove: SUCCESS
```

Missing operations during gap:
- `mcp__github-write__create_pull_request`
- `mcp__github-read__get_pull_request_files`
- Other GitHub API validation calls

## Solution

Add tool-level logging to GitHub MCP servers following existing pattern in `knowledge/procedures/mcp-tool-logging.md`.

Reference implementation: `atxtechbro-git-mcp-server` (already has logging)

## Principles This Supports

- **[Five Focusing Steps](knowledge/principles/five-focusing-steps.md)**: Can't identify the system constraint without visibility into all operations
- **[Selective Optimization](knowledge/principles/selective-optimization.md)**: Need timing data to prioritize highest value-to-effort ratio improvements
- **[Tracer Bullets](knowledge/principles/tracer-bullets.md)**: Feedback loops require observable units of change - can't adjust aim without seeing where shots land
- **[Systems Stewardship](knowledge/principles/systems-stewardship.md)**: Maintaining systems requires measurement and visibility into performance patterns
- **[Subtraction Creates Value](knowledge/principles/subtraction-creates-value.md)**: Cognitive buffer constraint requires knowing which tools consume the most tokens/time

## Expected Impact

With GitHub API timing visibility:
- Identify actual bottlenecks in `/close-issue` workflow
- Target scripting efforts at highest-impact operations
- Reduce 15-20 minute workflow to 5-10 minutes through selective optimization

## Acceptance Criteria

- [ ] GitHub MCP server calls appear in `~/mcp-tool-calls.log`
- [ ] Timestamps show duration of PR creation, issue updates, file validation
- [ ] Can identify which GitHub API calls are slowest/most frequent
- [ ] Update `knowledge/procedures/mcp-tool-logging.md` docs table to reflect new logging status

Start with knowledge/procedures/mcp-tool-logging.md and compare the reference implementation in atxtechbro-git-mcp-server. Locate the GitHub MCP server tool entry points, then verify that calls record timestamps and durations in ~/mcp-tool-calls.log. Update the documentation table and confirm that PR creation, issue updates, and file validation are visible.

Written by the indexing model from the issue text.