Add Muse Wonder to Meta Computer Use models

[NikhilShahi]

Summary

Support muse-wonder-1.0 in Meta Computer Use requests through the shared MetaComputerUseLlm literal used by TypedDict and legacy Pydantic requests.

Changes

• Add Wonder without changing request methods or defaults.
• Cover request serialization, response parsing, and sync/async request typing with Wonder.
• Bump the SDK to 1.11.1.

Validation

• poetry run ruff check . passed.
• poetry run pytest tests --ignore=tests/sandbox/e2e: 811 tests passed, including mypy/Pyright typing contracts.
• poetry check --lock and poetry build passed.
• Dictionary and legacy Python quick-start examples passed with mocked transport; no live agent task was run.

---

Note

Low Risk
Type literal and test-only updates with a patch version bump; no runtime behavior or API surface changes beyond allowing a new model string.

Overview
Adds muse-wonder-1.0 to the shared MetaComputerUseLlm literal so Meta Computer Use llm values are accepted by TypedDict and legacy Pydantic start params without changing client methods or defaults.

Tests are extended with parametrized coverage for Wonder alongside existing Spark models (response parsing, request serialization), and sync/async typecheck fixtures now exercise muse-wonder-1.0. SDK version is bumped to 1.11.1.

^{Reviewed by Cursor Bugbot for commit 6300aa8. Bugbot is set up for automated code reviews on this repo. Configure here.}

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring