Disable avoid-non-const-global-variables lint check (pytorch#62008)

Summary:
As GoogleTest `TEST` macro is non-compliant with it as well as `DEFINE_DISPATCH`

All changes but the ones to `.clang-tidy` are generated using following script:
```
for i in `find . -type f -iname "*.c*" -or -iname "*.h"|xargs grep cppcoreguidelines-avoid-non-const-global-variables|cut -f1 -d:|sort|uniq`;  do sed -i "/\/\/ NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)/d" $i; done
```

Pull Request resolved: pytorch#62008

Reviewed By: driazati, r-barnes

Differential Revision: D29838584

Pulled By: malfet

fbshipit-source-id: 1b2f8602c945bd4ce50a9bfdd204755556e31d13

Author: malfet

Diff for: .clang-tidy-oss

@@ -9,6 +9,7 @@ bugprone-*,
 -bugprone-reserved-identifier,
 cppcoreguidelines-*,
 -cppcoreguidelines-avoid-magic-numbers,
+-cppcoreguidelines-avoid-non-const-global-variables,
 -cppcoreguidelines-interfaces-global-init,
 -cppcoreguidelines-macro-usage,
 -cppcoreguidelines-owning-memory,

Diff for: aten/src/ATen/Context.cpp

@@ -235,7 +235,6 @@ Allocator* getCPUAllocator() {
 //    means the allow_tf32 flags are overrided and tf32 is force disabled
 // override_allow_tf32_flag = false
 //    means the original allow_tf32 flags are followed
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 thread_local bool override_allow_tf32_flag = false;
 
 NoTF32Guard::NoTF32Guard() {

Diff for: aten/src/ATen/MapAllocator.cpp

@@ -53,7 +53,6 @@ struct MapInfo {
   std::atomic<int> refcount;
 };
 
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 const std::string unknown_filename = "filename not specified";
 #ifdef _WIN32
 const std::string unknown_eventname = "eventname not specified";

Diff for: aten/src/ATen/ParallelNative.cpp

@@ -23,11 +23,9 @@ namespace at {
 namespace {
 // used with _set_in_parallel_region to mark master thread
 // as in parallel region while executing parallel primitives
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 thread_local bool in_parallel_region_ = false;
 
 // thread number (task_id) set by parallel primitive
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 thread_local int thread_num_ = 0;
 
 void _set_in_parallel_region(bool in_region) {
@@ -60,7 +58,6 @@ const int CONSUMED = -2;
 //  - NOT_SET - pool not initialized, user value is not set
 //  - positive value - pool not initialized, user value set
 //  - CONSUMED - pool is initialized
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 std::atomic<int> num_intraop_threads{NOT_SET};
 
 int _num_pool_threads(int nthreads) {

Diff for: aten/src/ATen/ParallelThreadPoolNative.cpp

@@ -17,7 +17,6 @@ const int CONSUMED = -2;
 // (CONSUMED - thread pool is initialized)
 // or
 // NOT_SET -> CONSUMED
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 std::atomic<int> num_interop_threads{NOT_SET};
 
 // thread pool global instance is hidden,
@@ -46,7 +45,6 @@ std::shared_ptr<TaskThreadPoolBase> create_c10_threadpool(
 
 } // namespace
 
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 C10_REGISTER_CREATOR(ThreadPoolRegistry, C10, create_c10_threadpool);
 
 void set_num_interop_threads(int nthreads) {

Diff for: aten/src/ATen/SequenceNumber.cpp

@@ -4,7 +4,6 @@ namespace at {
 namespace sequence_number {
 
 namespace {
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 thread_local uint64_t sequence_nr_ = 0;
 } // namespace
 

Diff for: aten/src/ATen/VmapMode.cpp

@@ -3,7 +3,6 @@
 namespace at {
 namespace impl {
 
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 thread_local int64_t VmapMode_current_vmap_level = 0;
 
 int64_t VmapMode::current_vmap_level() {

Diff for: aten/src/ATen/autocast_mode.cpp

@@ -47,14 +47,12 @@ namespace {
 // directly against incoming TensorImpl*s.
 using weakref_type = c10::weak_intrusive_ptr<TensorImpl, UndefinedTensorImpl>;
 using val_type = std::tuple<weakref_type, Tensor>;
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 thread_local std::unordered_map<TensorImpl*, val_type> cached_casts;
 
 // nesting tracks the nesting depth of the Python-side context manager.
 // When the autocast context manager exits to a nesting level that's outside
 // any instance of autocast (which should occur at the end of each forward pass)
 // it calls clear_cache() to ensure cached Tensors don't leak outside the autocasting region.
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 thread_local int nesting = 0;
 
 // autocast_cpu_dtype is the lower_precision_fp used by AutocastCPU.

Diff for: aten/src/ATen/benchmarks/quantize_per_channel.cpp

@@ -79,10 +79,7 @@ static void GenerateSizes2d(benchmark::internal::Benchmark* b) {
   }
 }
 
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 BENCHMARK(quantize_per_channel_2d)->Apply(GenerateSizes2d);
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 BENCHMARK(quantize_per_channel_4d_contiguous)->Apply(GenerateSizes4d);
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 BENCHMARK(quantize_per_channel_4d_channels_last)->Apply(GenerateSizes4d);
 BENCHMARK_MAIN();

Diff for: aten/src/ATen/benchmarks/stateful_conv1d.cpp

@@ -82,6 +82,5 @@ static void GenerateSizes(benchmark::internal::Benchmark* b) {
   }
 }
 
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 BENCHMARK(stateful_conv1d)->Apply(GenerateSizes);
 BENCHMARK_MAIN();

Diff for: aten/src/ATen/benchmarks/tensor_add.cpp

@@ -26,6 +26,5 @@ static void GenerateSizes(benchmark::internal::Benchmark* b) {
   }
 }
 
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 BENCHMARK(tensor_add)->Apply(GenerateSizes);
 BENCHMARK_MAIN();

Diff for: aten/src/ATen/core/Dimname.cpp

@@ -4,7 +4,6 @@
 
 namespace at {
 
-// NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)
 static Symbol kWildcard = Symbol::dimname("*");
 
 std::ostream& operator<<(std::ostream& out, const Dimname& dimname) {