fix: add local-model concurrency guard and expand model tests

[sanikolaev]

• Serialize causal model predictions to avoid concurrent Qwen crashes
• Add FFI concurrency tests for Qwen and other local models
• Auto-download required models for integration tests
• Document test cache env vars in README

Related issue #133

[github-actions]

Linux debug test results

  8 files    8 suites   13m 12s ⏱️
500 tests 480 ✅ 20 💤 0 ❌
514 runs  494 ✅ 20 💤 0 ❌

Results for commit b8efe7e.

♻️ This comment has been updated with latest results.

[github-actions]

Windows test results

  5 files    5 suites   17m 33s ⏱️
481 tests 468 ✅ 13 💤 0 ❌
489 runs  476 ✅ 13 💤 0 ❌

Results for commit b8efe7e.

♻️ This comment has been updated with latest results.

[github-actions]

Linux release test results

  8 files    8 suites   6m 29s ⏱️
500 tests 487 ✅ 13 💤 0 ❌
514 runs  501 ✅ 13 💤 0 ❌

Results for commit b8efe7e.

♻️ This comment has been updated with latest results.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ sanikolaev
❌ donhardman
_{You have signed the CLA already but the status is still pending? Let us recheck it.}