Add force think tag stripping to fix blank code boxes appearing while subagents are running

[Barragek0]

📝 What does this change?

Force think-tag stripping when tools are present in the request (subagent/tool-call invocations). This prevents tags in content from rendering as blank code blocks in the chat UI when using a subagent, flooding the chat with blank code boxes.

The trigger is tool-call detection (options.tools.length > 0), not model-specific. The ThinkTagFilter is a no-op for models that don't emit think tags, so this is safe for all models.

🧪 How did you test it?

Tested with MiMo v2.5

✅ Checklist

• npm run compile passes
• npm run lint passes
• npm test passes
• npm run package produces a VSIX
• I tested it works
• I updated docs/CHANGELOG if needed

[ltmoerdani]

Thanks for this one @Barragek0, clean approach. Reusing the existing stripThinkTags plumbing and forcing the mode when tools are present beats a model-specific hack, and the no-op passthrough for models that don't emit think tags makes it safe across the board.

I pulled the branch and verified beyond lint: the full npm run lint gate passes (all 7 checks), and I ran a direct logic check on createThinkTagFilter. forceOverride activates the filter even on never, non-forced requests keep their old behavior, and the visible/thinking split comes through intact. Also like that the stripped thinking still surfaces via LanguageModelThinkingPart, so nothing gets dropped.

One small follow-up I'll handle myself after merging: a tiny unit test for the force path, since thinkTags.ts has no coverage yet. Not a blocker.

Merging now. Thanks again!