[LLVMGPU] VectorDistribution pipeline for attention

[Groverkss]

This patch adds support for lowering attention through VectorDistribution pipeline. Currently, it has the following limitations which will be fixed as followups:

• Only 1 subgroup is used
• There are 4 shared memory promotions. Based on the intrinsic, this can be reduced to 2 or 3.

Depends on https://github.com/iree-org/iree/pull/17744

[Groverkss]

Please only review the last commit. Other commits are the patches this patch depends on.

[MaheshRavishankar]

I dont have any comments on this, skimming through this looks OK to me to land and iterate

[Groverkss]

Closing in favour of https://github.com/iree-org/iree/pull/17773