v08.11

Enhancements:

• Added initial MLX backend support including random.default_rng, dtype-aware creation wrappers, to_numpy, count_nonzero, ravel, and linalg.svd support.
• Added infer_backend_device_dtype for shared backend, device, and dtype inference from like values.
• Exported DoFunc for reusable call-time auto-dispatch, with faster do and get_namespace dispatch paths.
• Refreshed package metadata, pixi environments, docs infrastructure, and CI, including the testmlx environment and MLX CI job.
• Reworked backend tests around the unified XFAILS registry and expanded linear algebra coverage.

Bug Fixes:

• Cached failed .device and .dtype probes in get_namespace to avoid repeated slow exceptions for array classes such as JAX tracers (#30).
• Fixed autojit with lazy lazy_astype dtype evaluation.
• Added TensorFlow astype handling and a NumPy-like Torch nonzero wrapper.
• Fixed reduced and batched SVD shape handling in svd_manual_full_matrices_kwarg.
• Fixed prime_factors to preserve integer factors.

Full Changelog: v0.8.10...v0.8.11