Raise DecodeError for invalid UTF-8 TOML buffers - #1270

msgspec.toml.decode(b"name = '\xff'") currently raises a bare UnicodeDecodeError. Applications catching msgspec.DecodeError for malformed TOML do not catch it, even though syntax failures are already wrapped as DecodeError.

Keep UTF-8 conversion inside the parse-error boundary and wrap UnicodeDecodeError using the TOML wrapper's existing error convention. Buffer-type errors still propagate as TypeError. Regressions cover bytes, bytearray, and memoryview, with valid non-ASCII TOML controls.

Validation: all three malformed UTF-8 cases fail before the fix; the complete unit suite passes afterward (6553 passed, 116 existing skips). Ruff 0.17.0 lint and format checks pass for both changed files.