Add --merge and --prefix options for multi-machine archive management

[dpatou]

Summary

This PR adds two new options to the all command to support managing transcript archives across multiple machines:

• --merge / -m: Preserves existing sessions in the output directory when regenerating. Sessions from the source are regenerated, while sessions only present in the existing archive are kept in the index.

• --prefix: Adds a prefix label to sessions (displayed as [prefix] in the index). Useful for identifying which machine generated each session.

Use case

When using Claude Code on multiple machines, each machine has its own ~/.claude/projects/ folder. Without --merge, regenerating the archive from one machine would overwrite sessions from other machines.

With these options, you can:

# On machine A
claude-code-transcripts all --merge --prefix=laptop -o /shared/archive/

# On machine B  
claude-code-transcripts all --merge --prefix=desktop -o /shared/archive/

Each run preserves sessions from other machines while regenerating local sessions with a machine identifier.

Changes

• Added find_existing_sessions() function to parse existing archive's project_index.html
• Added merge_sessions() function to combine source and existing sessions
• Modified generate_batch_html() to support merge and prefix parameters
• Updated project_index.html template to display prefix labels
• Added comprehensive tests for all new functionality
• Updated README.md with documentation

Test plan

• Unit tests pass (uv run pytest)
• Manual testing with real session files
• Verified prefix display in generated HTML
• Verified merge preserves orphaned sessions