Next fix for gemm-allreduce two-shot

[aleozlx]

Previously:
#2171
NVIDIA/cutlass#2845

Next:
Update according to
https://github.com/NVIDIA/cutlass/blob/main/examples/python/CuTeDSL/distributed/distributed_gemm_all_reduce_blackwell.py

[aleozlx]

Currently not prioritizing this, as no blocker exists for anyone

[kaushik-rohit]

Hi, I have some time on my hand and I am looking to get into kernels. Can I pick this?

[aleozlx]

@kaushik-rohit of course! thanks. lmk if you have questions