Minor Cutlass change to fix CI (pytorch#863)

Summary:
Pull Request resolved: facebookresearch/FBGEMM#863

In the ongoing saga of updating cutlass to version 3.8V2, there remains a minor bug in the OSS main branch that is causing build failures in OSS CI. This small diff moves our cutlass dependency back to my fork which tracks cutlass main but with the fix added. Hopefully this resolves the GPU build issues we're currently seeing.

X-link: pytorch#3779

Reviewed By: q10

Differential Revision: D70734697

Pulled By: jwfromm

fbshipit-source-id: 0a22c426cfeaffa70725d80dacaf69d9f9e43fc1

Author: jwfromm

.gitmodules

@@ -14,7 +14,8 @@
 # Go back to main cutlass when possible.
 [submodule "external/cutlass"]
 	path = external/cutlass
-	url = https://github.com/NVIDIA/cutlass
+	url = https://github.com/jwfromm/cutlass
+	branch = FBGEMM
 [submodule "external/json"]
 	path = external/json
 	url = https://github.com/nlohmann/json.git