[MPS] `torch.mm` raises opaque function-state RuntimeError for `bool` matrices

### 🐛 Describe the bug

# [MPS] `torch.mm` raises opaque function-state RuntimeError for `bool` matrices

## Description

`torch.mm` on MPS raises an internal `RuntimeError: Failed to create function state object for: matmul_bool` when both operands have dtype `torch.bool`. The same call on CPU is rejected cleanly with `NotImplementedError: "addmm_impl_cpu_" not implemented for 'Bool'`.

## API under test

`torch.mm(input, mat2, *, out=None)`

## Broken convention

`torch.mm` performs matrix multiplication and expects numeric dtypes suitable for matmul / `addmm`. Boolean tensors are outside the supported domain for this op.

- Valid: floating and (where implemented) integer numeric dtypes with compatible 2D shapes `(n, m) @ (m, p)`
- Invalid: `torch.bool` (no boolean matrix-multiply kernel)

CPU enforces this with a clear dtype rejection. MPS does not guard at dispatch and instead routes the call into a Metal kernel path (`matmul_bool`) that fails during function-state initialization.

## To Reproduce

```python
import torch

a = torch.randint(0, 2, (4, 4), dtype=torch.bool, device="mps")
b = torch.randint(0, 2, (4, 4), dtype=torch.bool, device="mps")
torch.mm(a, b)
```

## Actual behavior

RuntimeError: Failed to create function state object for: matmul_bool

## Expected behavior

Raise a clear `NotImplementedError` (or equivalent) stating that `bool` is unsupported for `mm` on MPS, matching CPU:

NotImplementedError: "addmm_impl_cpu_" not implemented for 'Bool'

## Environment

- PyTorch version: 2.8.0
- OS: macOS 26.5 arm64
- Device: Apple Silicon (MPS)

### Versions

```
PyTorch version: 2.12.0
Is debug build: False
CUDA used to build PyTorch: None
ROCM used to build PyTorch: N/A

OS: macOS 26.5 (arm64)
GCC version: Could not collect
Clang version: 21.0.0 (clang-2100.1.1.101)
CMake version: version 4.3.4
Libc version: N/A

Python version: 3.14.5 (main, May 10 2026, 19:20:57) [Clang 22.1.3 ] (64-bit runtime)
Python platform: macOS-26.5-arm64-arm-64bit-Mach-O
Is CUDA available: False
CUDA runtime version: No CUDA
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
Is XPU available: False
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True
Caching allocator config: N/A

CPU:
Apple M5

Versions of relevant libraries:
[pip3] numpy==2.4.6
[pip3] torch==2.12.0
[conda] Could not collect
```

cc @malfet @kulinseth @DenisVieriu97 @jhavukainen @aditvenk @Isalia20

The issue names no source file or test. Start by running the provided torch.mm reproduction on MPS and tracing the MPS dispatch path for boolean matrices; done means the unsupported dtype produces a clear NotImplementedError or equivalent and a regression test covers the case.

Written by the indexing model from the issue text.