[feat] Sharding logic split to pattern detection and executor for EP and BMM (fixes NVIDIA#5916) (#94)

* Updated tests

Signed-off-by: greg-kwasniewski1 <[email protected]>

* fixed tp sharding bug

Signed-off-by: greg-kwasniewski1 <[email protected]>

* Fixed sharding tests

Signed-off-by: greg-kwasniewski1 <[email protected]>

* Fixed sharding tests 1.1

Signed-off-by: greg-kwasniewski1 <[email protected]>

* import fix

Signed-off-by: Lucas Liebenwein <[email protected]>

---------

Signed-off-by: greg-kwasniewski1 <[email protected]>
Signed-off-by: Lucas Liebenwein <[email protected]>
Co-authored-by: Lucas Liebenwein <[email protected]>

Author: greg-kwasniewski1

tensorrt_llm/_torch/auto_deploy/transformations/library/__init__.py

@@ -3,7 +3,6 @@
 from .attention import *
 from .collectives import *
 from .eliminate_redundant_transposes import *
-from .ep_sharding import *
 from .fused_moe import *
 from .fusion import *
 from .kvcache import *