Add subgroup matrix multiplication

[junjihashimoto]

Add subgroup matrix multiplication.
The kernel can be executed with the subgroupMatrixMultiplication function, but the results are incorrect. I am currently debugging.

> sysctl -n machdep.cpu.brand_string
Apple M4 Max
> MATMUL_VERSION=12 ./build/matmul  | grep -A 2 'Dispatching\|Exec'
matmul(50174,0x2016ee140) malloc: nano zone abandoned due to inability to reserve vm space.
[info] Dispatching Kernel version 12: f16: Subgroup matrix multiply with transpose, 30 iterations ...
[info] Copying result to CPU
[info]
--
Execution Time: (M = 4096, K = 4096, N = 8192) x 30 iterations :
40.0 milliseconds / dispatch ~ 6871.66 GFLOPS
================================================================================
> MATMUL_VERSION=11 ./build/matmul  | grep -A 2 'Dispatching\|Exec'
matmul(13932,0x2016ee140) malloc: nano zone abandoned due to inability to reserve vm space.
[info] Dispatching Kernel version 11: f16: 2D blocktiling with loop unrolling, vectorization and transpose, 30 iterations ...
[info] Copying result to CPU
[info]
--
Execution Time: (M = 4096, K = 4096, N = 8192) x 30 iterations :
26.6 milliseconds / dispatch ~ 10316.68 GFLOPS ## This is the result not using the subgroupMatrixMultiplication function.
================================================================================

[junjihashimoto]

The main branch does not seem to output any shader compilation errors.