Integrate triton_kernels in sgl-kernel (#8762)

sgl-kernel/CMakeLists.txt

@@ -66,6 +66,16 @@ FetchContent_Declare(
     GIT_SHALLOW    OFF
 )
 FetchContent_Populate(repo-deepgemm)
+
+# Triton
+FetchContent_Declare(
+    repo-triton
+    GIT_REPOSITORY "https://github.com/triton-lang/triton"
+    GIT_TAG        8f9f695ea8fde23a0c7c88e4ab256634ca27789f
+    GIT_SHALLOW    OFF
+)
+FetchContent_Populate(repo-triton)
+
 # flashinfer
 FetchContent_Declare(
     repo-flashinfer
@@ -413,3 +423,9 @@ install(DIRECTORY "${repo-cutlass_SOURCE_DIR}/include/cute/"
 
 install(DIRECTORY "${repo-cutlass_SOURCE_DIR}/include/cutlass/"
         DESTINATION "deep_gemm/include/cutlass")
+
+# triton_kernels
+install(DIRECTORY "${repo-triton_SOURCE_DIR}/python/triton_kernels/triton_kernels/"
+        DESTINATION "triton_kernels"
+        PATTERN ".git*" EXCLUDE
+        PATTERN "__pycache__" EXCLUDE)