feat: Add appropriate "think" trigger levels to .claude/commands

## Context
The Claude model has an undocumented "extended thinking mode" triggered by specific keywords that allocate progressively more computational budget:
- `think` - baseline extended thinking
- `think hard` - increased budget
- `think harder` - more budget
- `ultrathink` - maximum budget

## Problem
This is currently an "easter egg" feature that creates Claude-specific vendor lock-in. However, since our .claude/commands are already Claude-specific, we should strategically incorporate these triggers where they add value.

## Proposed Solution
Review existing commands in `.claude/commands/` and add appropriate thinking levels where beneficial. Key considerations:

1. **Not all commands need extended thinking** - More thinking isn't always better
2. **Match thinking level to task complexity**:
- `/retro` - Could benefit from `think hard` for deeper analysis
- Complex debugging commands - May warrant `think harder`
- Simple commands - No thinking trigger needed

## Tasks
- [ ] Audit all existing .claude/commands
- [ ] Identify commands that would benefit from extended thinking
- [ ] Add appropriate thinking triggers based on task complexity
- [ ] Document the thinking level choice rationale in comments
- [ ] Test to ensure thinking triggers improve output quality

## Notes
- This creates intentional Claude dependency, but commands are already in `.claude/` directory
- Should be selective - not every command needs extended thinking
- Consider user wait time vs. output quality tradeoff

Start by auditing the existing files under `.claude/commands/` and classify each command by task complexity. Select appropriate thinking triggers, document the rationale in comments, and test whether the chosen levels improve output quality without unnecessary wait time.

Written by the indexing model from the issue text.