add dsv3 mi300 triton config for block scale

[BruceXcluding]

Motivation

Tune MoE:

python benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py --model deepseek-ai/DeepSeek-V3 --tp-size 8 --dtype fp8_w8a8 --tune

The offline perf would get a 20% uplift with configs. server perf no obvious change.

Modifications

Checklist

• Format your code according to the Code Formatting with Pre-Commit.
• Add unit tests as outlined in the Running Unit Tests.
• Update documentation / docstrings / example tutorials as needed, according to Writing Documentation.
• Provide throughput / latency benchmark results and accuracy evaluation results as needed, according to Benchmark and Profiling.

[BruceXcluding]

@HaiShaw Please check

[HaiShaw]

Maybe fix following text/example usage to what is to be in main?

Gemm Tune with private repo
export TRITON_PRINT_AUTOTUNING=1
python tune_gemm.py

[HaiShaw]

LGTM after minors addressed, thanks!

[BruceXcluding]

Maybe fix following text/example usage to what is to be in main?

Gemm Tune with private repo
export TRITON_PRINT_AUTOTUNING=1
python tune_gemm.py

I'll remove this part of the description. If a tune_gemm example is added in the main, I will add rocm configs.

[merrymercy]

ready to be merged?