Better Grouped GEMM + EP (#45621)
* init
* style
* full support
* support EP better using offsets !
* comments
* get rid of neutralize_ep_sentinels
* remove deepgemm stuff
* fix
* prefix
* move
* fix
* remove comment
* fix unintilized outputs leaking
* revert unnecessary changes
* more unnecessary changes
* revert downcast
* keep it simple
* guard deepgemm cuda version
* fix style
* moe sentinel support
* fix
* compilable sonicmoe
* fix
* dtensor support
* more dtensor
* simpler
* remove comment
* make loaders return a dataclass/namespace that's correctly typed
* style
* fix
