[GPU][Codegen] Implement virtual dense mfma (VDMFMA) (#23677)

This PR introduces virtual dense MFMAs (VDMFMA), which implement the
sparse trick for skinny GEMM (M=8) workloads inspired by the hugging
face article [Creating custom kernels for the AMD
MI300](https://huggingface.co/blog/mi300kernels). This patch implements
VDMFMA for all CDNA3 variants of the sparse mfma intrinsic. Further
plumbing and support for CDNA4 variants will be future PRs.

Sparse MFMA (V_SMFMAC) instructions perform MMA on an imbalanced pair of
operands: a 4:2 structured-sparse matrix A and a dense matrix B. The
instruction also takes a sparsity index that encodes which 2 of every 4
elements along K are non-zero within the sparse matrix A. The trick
exploits this by pairing even/odd lanes to jointly describe a full dense
row.

The benefit over the current approach of padding M=8 to M=16 for dense
MFMA is that sparse MFMA processes twice the K-depth of the equivalent
dense MFMA in the same number of cycles.

Performance comparison for FP16/FP8 (coupled with optimizing out
redundant accumulator processing):

```
shape              vdmfma  baseline  %
-----------------  ------  --------  -----
f16_8x13312x16384  214 us  217 us    +1.4%
f16_8x13312x8192   123 us  122 us    -
f16_8x2304x16384   137 us  145 us    +5.5%
f16_8x2304x8192    114 us  115 us    -
f16_8x6656x16384   144 us  145 us    -
f16_8x6656x8192    116 us  123 us    +5.7%
fp8_8x13312x16384  135 us  133 us    -
fp8_8x13312x8192   113 us  106 us    -6.6%
fp8_8x2304x16384   113 us  126 us    +10.3%
fp8_8x2304x8192    92.6 us 98.6 us   +6.1%
fp8_8x6656x16384   114 us  133 us    +14.3%
fp8_8x6656x8192    97.8 us 108 us    +9.4%
```
Assisted by: Claude

---------

Signed-off-by: Eric Feng <Eric.Feng@amd.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: efric