[agent-sdk-j2s] agent-server runtime LLM switching

[enyst]

Implements runtime LLM switching end-to-end for remote clients (VS Code extension) without agent immutability/diff enforcement.

Key changes:

• Agent is componentized: restore-time + runtime swaps (LLM, tools, context) are allowed; no diff-based enforcement.
• Conversation persistence remains durable (base_state + events); LLM selection persists across restart; runtime secrets override persisted secrets on restore.
• Agent-server REST: POST /api/conversations/{id}/llm (profile_id or inline llm) and POST /api/conversations/{id}/llm/switch.
• Tests: agent-server wsproto integration covers large event history + restore + runtime LLM switching.

Docs: docs/llm_switching.md
Example: examples/02_remote_agent_server/07_llm_switch_and_restore.py

Docs: check-examples uses matching OpenHands/docs branches; docs PR OpenHands/docs#219 is optional follow-up for landing docs on main.
Beads: agent-sdk-j2s

[enyst]

Rebased onto updated agent-sdk-26-llm-switch (which now includes latest upstream/main via profile branch). Only remaining failing check is check-examples, pending OpenHands/docs#219.

[all-hands-bot]

[Automatic Post]: It has been a while since there was any activity on this PR. @enyst, are you still working on it? If so, please go ahead, if not then please request review, close it, or request that someone else follow up.

[all-hands-bot]

[Automatic Post]: It has been a while since there was any activity on this PR. @enyst, are you still working on it? If so, please go ahead, if not then please request review, close it, or request that someone else follow up.