Address all review feedback: OpenAI-format current.jsonl, unsummarized.jsonl, keep_recent_messages config, file concatenation, queue UUID assignment

- Change current.jsonl to OpenAI API format (no seq/timestamp/tokens fields)
- Add unsummarized.jsonl for keeping recent N messages during compression
- Add keep_recent_messages config (default 5)
- Update compression to extract last N messages to unsummarized.jsonl
- Change get_current_context_file() to accept unsummarized_file_path parameter
- Update recreate_current_context() to accept unsummarized_file_path
- Specify UUID assignment in existing queue code (main.py fetch_and_enqueue_tasks)
- Update TaskFactory.from_dict() to set UUID from task dict
- Change task.get_uuid() to task.uuid attribute access
- Remove all code examples, keep only Japanese specifications
- Update all LLMClients to use file concatenation for request.json (no memory loading)
- Specify request.json format varies by LLM type
- Add cleanup_days to config examples throughout
- Fix all file section numbers (3.4->3.5->3.6->3.7)

Co-authored-by: notfolder <20558197+notfolder@users.noreply.github.com>

Author: Copilot