Build / CMake can't find nvcc

Describe the bug
I'm trying to build the ONNX Runtime on Ubuntu 20.04, with CUDA 11.4 installed from the https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64 apt repo, but it fails to find nvcc, even though I'm passing --cuda_home

-- 
-- ******** Summary ********
--   CMake version             : 3.21.2
--   CMake command             : /home/spike/Applications/cmake/bin/cmake
--   System                    : Linux
--   C++ compiler              : /usr/bin/c++
--   C++ compiler version      : 9.3.0
--   CXX flags                 :  -ffunction-sections -fdata-sections -DCPUINFO_SUPPORTED -Wnon-virtual-dtor
--   Build type                : RelWithDebInfo
--   Compile definitions       : ENABLE_ORT_FORMAT_LOAD;ORT_ENABLE_ORT_FORMAT_RUNTIME_GRAPH_OPTIMIZATION;EIGEN_MPL2_ONLY;PLATFORM_POSIX
--   CMAKE_PREFIX_PATH         : 
--   CMAKE_INSTALL_PREFIX      : /usr/local
--   CMAKE_MODULE_PATH         : /home/spike/repos/onnxruntime/cmake/external
-- 
--   ONNX version              : 1.10.1
--   ONNX NAMESPACE            : onnx
--   ONNX_USE_LITE_PROTO       : ON
--   USE_PROTOBUF_SHARED_LIBS  : OFF
--   Protobuf_USE_STATIC_LIBS  : ON
--   ONNX_DISABLE_EXCEPTIONS   : OFF
--   ONNX_WERROR               : OFF
--   ONNX_BUILD_TESTS          : OFF
--   ONNX_BUILD_BENCHMARKS     : OFF
--   ONNXIFI_DUMMY_BACKEND     : OFF
--   ONNXIFI_ENABLE_EXT        : OFF
-- 
--   Protobuf compiler         : 
--   Protobuf includes         : 
--   Protobuf libraries        : 
--   BUILD_ONNX_PYTHON         : OFF
CMake Error at /home/spike/Applications/cmake/share/cmake-3.21/Modules/CMakeDetermineCompilerId.cmake:741 (message):
  Compiling the CUDA compiler identification source file
  "CMakeCUDACompilerId.cu" failed.

  Compiler: /usr/bin/nvcc

  Build flags:

  Id flags: -v

  

  The output was:

  No such file or directory

  

  

Call Stack (most recent call first):
  /home/spike/Applications/cmake/share/cmake-3.21/Modules/CMakeDetermineCompilerId.cmake:6 (CMAKE_DETERMINE_COMPILER_ID_BUILD)
  /home/spike/Applications/cmake/share/cmake-3.21/Modules/CMakeDetermineCompilerId.cmake:48 (__determine_compiler_id_test)
  /home/spike/Applications/cmake/share/cmake-3.21/Modules/CMakeDetermineCUDACompiler.cmake:283 (CMAKE_DETERMINE_COMPILER_ID)
  CMakeLists.txt:1465 (enable_language)

Urgency
Blocked

System information

• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Ubuntun 20.04
• ONNX Runtime installed from (source or binary): source
• ONNX Runtime version: 1.8.2
• Python version: 3.8.10
• Visual Studio version (if applicable):
• GCC/Compiler version (if compiling from source): Ubuntu 9.3.0-17ubuntu1~20.04
• CUDA/cuDNN version: 11.4 / 8.2.2
• GPU model and memory: RTX 2080 Ti

To Reproduce
Installed cmake to /home/spike/Applications/cmake and have /home/spike/Applications/cmake/bin on my path.
Added /usr/local/cuda/bin to my path, since the cuda package doesn't add nvcc it to /usr/bin

$ which nvcc
/usr/local/cuda/bin/nvcc

Building with

./build.sh --cuda_home /usr/local/cuda --cudnn_home /usr/lib/x86_64-linux-gnu/ --use_cuda --config RelWithDebInfo --build_shared_lib --parallel

Expected behavior
Builds and finds CUDA install at /usr/local/cuda

Start with build.sh and CMakeLists.txt around line 1465, then inspect the CMake compiler-identification failure shown in the report. Reproduce the CUDA build using the stated Ubuntu, CMake, CUDA, and --cuda_home settings, and confirm that the build locates nvcc at /usr/local/cuda/bin and proceeds past compiler detection.