Context checkpoint: before exploratory work; later `rewind`, retaining only concise report.

Use for investigations with many intermediate tool calls (`read`/`grep`/`glob`/`lsp`/etc.) to minimize subsequent context cost.

Rules:
- MUST `rewind` before yielding after starting a checkpoint.
- NEVER `checkpoint` while another checkpoint active.
- Subagents: disabled by default. Enable: agent-definition `tools:` frontmatter lists `checkpoint` or `rewind`; sister tool auto-included; requires `checkpoint.enabled` setting.

Typical flow:
1. `checkpoint(goal: …)`
2. Exploratory work
3. `rewind(report: …)` with concise findings

After `rewind`: intermediate checkpoint messages removed from active context; replaced by report.
