Rewrite C++ API tests in gtest (pytorch#11953)

Summary:
This PR is a large codemod to rewrite all C++ API tests with GoogleTest (gtest) instead of Catch.

You can largely trust me to have correctly code-modded the tests, so it's not required to review every of the 2000+ changed lines. However, additional things I changed were:

1. Moved the cmake parts for these tests into their own `CMakeLists.txt` under `test/cpp/api` and calling `add_subdirectory` from `torch/CMakeLists.txt`
2. Fixing DataParallel tests which weren't being compiled because `USE_CUDA` wasn't correctly being set at all.
3. Updated README

ezyang ebetica
Pull Request resolved: pytorch#11953

Differential Revision: D9998883

Pulled By: goldsborough

fbshipit-source-id: affe3f320b0ca63e7e0019926a59076bb943db80

Author: goldsborough

test/cpp/api/CMakeLists.txt

@@ -0,0 +1,56 @@
+set(TORCH_API_TEST_DIR "${TORCH_ROOT}/test/cpp/api")
+set(TORCH_API_TEST_SOURCES
+  ${TORCH_API_TEST_DIR}/any.cpp
+  ${TORCH_API_TEST_DIR}/cursor.cpp
+  ${TORCH_API_TEST_DIR}/expanding-array.cpp
+  ${TORCH_API_TEST_DIR}/integration.cpp
+  ${TORCH_API_TEST_DIR}/jit.cpp
+  ${TORCH_API_TEST_DIR}/main.cpp
+  ${TORCH_API_TEST_DIR}/memory.cpp
+  ${TORCH_API_TEST_DIR}/misc.cpp
+  ${TORCH_API_TEST_DIR}/module.cpp
+  ${TORCH_API_TEST_DIR}/modules.cpp
+  ${TORCH_API_TEST_DIR}/optim.cpp
+  ${TORCH_API_TEST_DIR}/ordered-dict.cpp
+  ${TORCH_API_TEST_DIR}/rnn.cpp
+  ${TORCH_API_TEST_DIR}/sequential.cpp
+  ${TORCH_API_TEST_DIR}/serialize.cpp
+  ${TORCH_API_TEST_DIR}/static.cpp
+  ${TORCH_API_TEST_DIR}/tensor_cuda.cpp
+  ${TORCH_API_TEST_DIR}/tensor_options_cuda.cpp
+  ${TORCH_API_TEST_DIR}/tensor_options.cpp
+  ${TORCH_API_TEST_DIR}/tensor.cpp
+)
+
+if (USE_CUDA)
+  list(APPEND TORCH_API_TEST_SOURCES ${TORCH_API_TEST_DIR}/parallel.cpp)
+endif()
+
+add_executable(test_api ${TORCH_API_TEST_SOURCES})
+target_include_directories(test_api PRIVATE ${ATen_CPU_INCLUDE})
+target_link_libraries(test_api PRIVATE torch gtest)
+
+if (USE_CUDA)
+  target_link_libraries(test_api PRIVATE
+    ${CUDA_LIBRARIES}
+    ${CUDA_NVRTC_LIB}
+    ${CUDA_CUDA_LIB}
+    ${TORCH_CUDA_LIBRARIES})
+  target_compile_definitions(test_api PRIVATE "USE_CUDA")
+endif()
+
+if (NOT MSVC)
+  if (APPLE)
+    target_compile_options(test_api PRIVATE
+      # Clang has an unfixed bug leading to spurious missing braces
+      # warnings, see https://bugs.llvm.org/show_bug.cgi?id=21629
+      -Wno-missing-braces)
+else()
+    target_compile_options(test_api PRIVATE
+      # Considered to be flaky.  See the discussion at
+      # https://github.com/pytorch/pytorch/pull/9608
+      -Wno-maybe-uninitialized
+      # gcc gives nonsensical warnings about variadic.h
+      -Wno-unused-but-set-parameter)
+  endif()
+endif()

test/cpp/api/README.md

@@ -1,28 +1,35 @@
-# C++ API Tests
+# C++ Frontend Tests
 
-In this folder live the tests for PyTorch's C++ API (formerly known as
-autogradpp). They use the [Catch2](https://github.com/catchorg/Catch2) test
-framework.
+In this folder live the tests for PyTorch's C++ Frontend. They use the
+[GoogleTest](https://github.com/google/googletest) test framework.
 
 ## CUDA Tests
 
-The way we handle CUDA tests is by separating them into a separate `TEST_CASE`
-(e.g. we have `optim` and `optim_cuda` test cases in `optim.cpp`), and giving
-them the `[cuda]` tag. Then, inside `main.cpp` we detect at runtime whether
-CUDA is available. If not, we disable these CUDA tests by appending `~[cuda]`
-to the test specifications. The `~` disables the tag.
+To make a test runnable only on platforms with CUDA, you should suffix your
+test with `_CUDA`, e.g.
 
-One annoying aspect is that Catch only allows filtering on test cases and not
-sections. Ideally, one could have a section like `LSTM` inside the `RNN` test
-case, and give this section a `[cuda]` tag to only run it when CUDA is
-available. Instead, we have to create a whole separate `RNN_cuda` test case and
-put all these CUDA sections in there.
+```cpp
+TEST(MyTestSuite, MyTestCase_CUDA) { }
+```
+
+To make it runnable only on platforms with at least two CUDA machines, suffix
+it with `_MultiCUDA` instead of `_CUDA`, e.g.
+
+```cpp
+TEST(MyTestSuite, MyTestCase_MultiCUDA) { }
+```
+
+There is logic in `main.cpp` that detects the availability and number of CUDA
+devices and supplies the appropriate negative filters to GoogleTest.
 
 ## Integration Tests
 
 Integration tests use the MNIST dataset. You must download it by running the
 following command from the PyTorch root folder:
 
-```shell
+```sh
 $ python tools/download_mnist.py -d test/cpp/api/mnist
 ```
+
+The required paths will be referenced as `test/cpp/api/mnist/...` in the test
+code, so you *must* run the integration tests from the PyTorch root folder.