Python: add double-buffer context window reducer for chat history

[marklubin]

Motivation and Context

Current chat history reduction strategies use stop-the-world summarization when context fills. This PR adds a ChatHistoryDoubleBufferReducer that proactively manages context through double buffering — beginning checkpoint summarization at a configurable threshold while the agent continues working, then swapping to the pre-built back buffer seamlessly.

Reference: https://marklubin.me/posts/hopping-context-windows/

Description

Three-phase algorithm:

1. Checkpoint (Phase 1): At checkpoint_threshold (default 0.7), fire off background summarization via asyncio.create_task to seed the back buffer. The agent continues working immediately.
2. Concurrent (Phase 2): New messages go to both active and back buffers via add_message_async.
3. Swap (Phase 3): At swap_threshold (default 0.95), swap to back buffer. If checkpoint isn't done yet, block with configurable timeout (default 120s). If no back buffer exists, fall back to synchronous checkpoint (never skip compaction).

Incremental summary accumulation across generations with max_generations control and renewal policies (RECURSE for meta-summarization, DUMP for clean restart).

Contribution Checklist

• Code builds clean without errors or warnings
• Follows SK Contribution Guidelines
• All unit tests pass (19/19); new tests added
• Did not break any existing functionality

AI Disclosure

This contribution was developed with assistance from Claude (Anthropic). All code was reviewed, tested, and validated by the author.

[marklubin]

@marklubin please read the following Contributor License Agreement(CLA). If you agree with the CLA, please reply with the following information.

@microsoft-github-policy-service agree [company="{your company}"]

Options:

• (default - no company specified) I have sole ownership of intellectual property rights to my Submissions and I am not making Submissions in the course of work for my employer.

@microsoft-github-policy-service agree

• (when company given) I am making Submissions in the course of work for my employer (or my employer has intellectual property rights in my Submissions by contract or applicable law). I have permission from my employer to make Submissions and enter into this Agreement on behalf of my employer. By signing below, the defined term “You” includes me and my employer.

@microsoft-github-policy-service agree company="Microsoft"

Contributor License Agreement

@microsoft-github-policy-service agree