[BUG] Possible Bug when exporting Kernel as PyTorch extension

I have the following script:

`import random
import cutlass
import torch

print_module = False

batch = 256
feature_dim_in = 4098
feature_dim_out = 10

type_A = torch.int8
type_B = torch.int8
type_C = torch.int32
type_D = torch.int32

torch.random.manual_seed(12)
random.seed(22)

input = torch.randint(-3, 3, size=(batch,feature_dim_in), device='cuda').to(type_A)
weight = torch.randint(-3, 3, size=(feature_dim_out,feature_dim_in), device='cuda').to(type_B)
weight_col_major = weight.t().contiguous()#.t()
bias = torch.randint(-3, 3, size=(batch,feature_dim_out), device='cuda').to(type_C)

result_tensor = torch.zeros_like(bias)
#Forward pass:

plan = cutlass.op.Gemm(
alpha=1, beta=1,
element_A=type_A,
element_B=type_B,
element_C=type_C,
element_D=type_D,
layout_A=cutlass.LayoutType.RowMajor,
layout_B=cutlass.LayoutType.RowMajor,
layout_C=cutlass.LayoutType.RowMajor
)
output_torch = (torch.matmul(input.float(),weight.t().float()) + bias.float()).to(torch.int32)
plan.opclass = cutlass.OpcodeClass.Simt
plan.run(input, weight_col_major, bias, result_tensor, print_module=print_module)

op = plan.construct()
linear_layer_gemm = cutlass.emit.pytorch(op, name='linear_layer', cc=plan.cc, sourcedir='linear', jit=True)`

when I call the python script I receive following error Traceback:

`Traceback (most recent call last):
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 2107, in _run_ninja_build
subprocess.run(
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/subprocess.py", line 516, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['ninja', '-v']' returned non-zero exit status 1.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "linear_layer.py", line 55, in
linear_layer_gemm = cutlass.emit.pytorch(op, name='linear_layer', cc=plan.cc, sourcedir='linear', jit=True)
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass/emit/pytorch.py", line 927, in pytorch
return _pytorch_gemm(device_op, name, cc, jit, sourcedir)
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass/emit/pytorch.py", line 775, in _pytorch_gemm
return _jit(name, cc, cpp_file, cuda_file)
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass/emit/pytorch.py", line 697, in jit
jitmodule = load(
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1309, in load
return jit_compile(
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1719, in jit_compile
write_ninja_file_and_build_library(
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1832, in write_ninja_file_and_build_library
run_ninja_build(
File "/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 2123, in run_ninja_build
raise RuntimeError(message) from e
RuntimeError: Error building extension 'linear_layer': [1/3] c++ -MMD -MF linear_layer.o.d -DTORCH_EXTENSION_NAME=linear_layer -DTORCH_API_INCLUDE_EXTENSION_H -DPYBIND11_COMPILER_TYPE="gcc" -DPYBIND11_STDLIB="libstdcpp" -DPYBIND11_BUILD_ABI="cxxabi1011" -I/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass_library/source/include -I/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass_library/source/tools/util/include -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/TH -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/THC -isystem /home/miniconda3/envs/cuda11-8/include -isystem /home/miniconda3/envs/cuda11-8/include/python3.8 -D_GLIBCXX_USE_CXX11_ABI=0 -fPIC -std=c++17 -c /home/reproai/reprai/extensions/linear/linear_layer.cpp -o linear_layer.o
[2/3] /home/miniconda3/envs/cuda11-8/bin/nvcc --generate-dependencies-with-compile --dependency-output linear_layer_kernel.cuda.o.d -DTORCH_EXTENSION_NAME=linear_layer -DTORCH_API_INCLUDE_EXTENSION_H -DPYBIND11_COMPILER_TYPE="gcc" -DPYBIND11_STDLIB="libstdcpp" -DPYBIND11_BUILD_ABI="cxxabi1011" -I/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass_library/source/include -I/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass_library/source/tools/util/include -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/TH -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/THC -isystem /home/miniconda3/envs/cuda11-8/include -isystem /home/miniconda3/envs/cuda11-8/include/python3.8 -D_GLIBCXX_USE_CXX11_ABI=0 -D__CUDA_NO_HALF_OPERATORS -D__CUDA_NO_HALF_CONVERSIONS -D__CUDA_NO_BFLOAT16_CONVERSIONS -D__CUDA_NO_HALF2_OPERATORS --expt-relaxed-constexpr -gencode=arch=compute_89,code=sm_89 --compiler-options '-fPIC' -std=c++17 -c /home/reproai/reprai/extensions/linear/linear_layer_kernel.cu -o linear_layer_kernel.cuda.o
FAILED: linear_layer_kernel.cuda.o
/home/miniconda3/envs/cuda11-8/bin/nvcc --generate-dependencies-with-compile --dependency-output linear_layer_kernel.cuda.o.d -DTORCH_EXTENSION_NAME=linear_layer -DTORCH_API_INCLUDE_EXTENSION_H -DPYBIND11_COMPILER_TYPE="gcc" -DPYBIND11_STDLIB="libstdcpp" -DPYBIND11_BUILD_ABI="cxxabi1011" -I/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass_library/source/include -I/home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/cutlass_library/source/tools/util/include -isystem /home/miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include -isystem /home//miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -isystem /home//miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/TH -isystem /home//miniconda3/envs/cuda11-8/lib/python3.8/site-packages/torch/include/THC -isystem /home//miniconda3/envs/cuda11-8/include -isystem /home//miniconda3/envs/cuda11-8/include/python3.8 -D_GLIBCXX_USE_CXX11_ABI=0 -D__CUDA_NO_HALF_OPERATORS -D__CUDA_NO_HALF_CONVERSIONS -D__CUDA_NO_BFLOAT16_CONVERSIONS -D__CUDA_NO_HALF2_OPERATORS --expt-relaxed-constexpr -gencode=arch=compute_89,code=sm_89 --compiler-options '-fPIC' -std=c++17 -c /home//reproai/reprai/extensions/linear/linear_layer_kernel.cu -o linear_layer_kernel.cuda.o
/home//reproai/reprai/extensions/linear/linear_layer_kernel.cu(106): error: namespace "torch" has no member "I32"

1 error detected in the compilation of "/home//reproai/reprai/extensions/linear/linear_layer_kernel.cu".
ninja: build stopped: subcommand failed.
`

Start with cutlass/emit/pytorch.py, especially the _pytorch_gemm and _jit paths named in the traceback, then inspect generated linear_layer_kernel.cu at line 106. Reproduce the supplied script and verify that the exported PyTorch extension compiles without the reported torch.I32 error.