Update sgl-kernel and remove fast hadamard depedency (#11844)

docker/Dockerfile

@@ -9,7 +9,7 @@ ARG DEEPEP_COMMIT=9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee
 ARG FLASHMLA_COMMIT=1408756a88e52a25196b759eaf8db89d2b51b5a1
 ARG FAST_HADAMARD_TRANSFORM_COMMIT=7fd811c2b47f63b0b08d2582619f939e14dad77c
 ARG CMAKE_BUILD_PARALLEL_LEVEL=2
-ARG SGL_KERNEL_VERSION=0.3.15
+ARG SGL_KERNEL_VERSION=0.3.16.post3
 ENV DEBIAN_FRONTEND=noninteractive \
     CUDA_HOME=/usr/local/cuda \
     GDRCOPY_HOME=/usr/src/gdrdrv-2.4.4/ \
@@ -152,14 +152,6 @@ RUN if [ "$TARGETARCH" = "amd64" ]; then \
       pip install -v . ; \
     fi
 
-# Install fast-hadamard-transform
-RUN if [ "$TARGETARCH" = "amd64" ]; then \
-      git clone https://github.com/Dao-AILab/fast-hadamard-transform && \
-      cd fast-hadamard-transform && \
-      git checkout ${FAST_HADAMARD_TRANSFORM_COMMIT} && \
-      pip install . ; \
-    fi
-
 # Python tools
 RUN python3 -m pip install --no-cache-dir \
     datamodel_code_generator \