Add spin_lock_atom_cas_acquire_wait function

[aleozlx]

For #2845

Added spin_lock_atom_cas_acquire_wait function to handle spin lock acquisition with atomic compare-and-swap.

[aleozlx]

This is functional. flashinfer-ai/flashinfer#2171

Raising it as a proposed solution for what we needed when upgrading to nvidia-cutlass-dsl 4.3.1 #2845

Kind regards from FlashInfer & cuDNN :)

[XiaoSong9905]

acquire wait is not needed. slack Xiao Song and we can schedule a meeting to explain this

[XiaoSong9905]

the two shot all redue.py fail is related to something else, let's discuss this in the meeting

[shubaoyu2]

you can use the new two-shot gemm+ar kernel in cutedsl examples. The one in flashinfer should be an old version.

adding something to CuTeDSL wheel package will take some time, so I would recommend you use the new kernel.

[aleozlx]

sounds good will discuss with you over slack. will learn about the new kernel example and bring action item back to FI