Make Transformers compilable by C++17 (pytorch#90389)

`register` keyword is removed in C++17, but keeping it there under ifdef
as I have not measured the perf implication on older compiler, though
there shouldn't be any: all modern compilers supposed to downright
ignore it.

This code originates from facebookresearch/xformers#375 will propose similar PR to remove register keyword usage to that repo.

Yet another thing discovered while working on pytorch#85969

Pull Request resolved: pytorch#90389
Approved by: https://github.com/drisspg

aten/src/ATen/native/transformers/cuda/mem_eff_attention/kernel_backward.h

@@ -687,7 +687,11 @@ struct AttentionBackwardKernel {
   static CUTLASS_DEVICE void kernel(Params& p_) {
     // Hint to nvcc to store points & tensor shapes in registers
     // as we use them a lot
+#if __cplusplus < 201703L
     register const Params p = p_;
+#else
+    const Params p = p_;
+#endif
 
     extern __shared__ char smem_buffer[];
     SharedStorage& shared_storage = *((SharedStorage*)smem_buffer);
@@ -721,7 +725,11 @@ struct AttentionBackwardKernel {
       __syncthreads();
     }
 
+#if __cplusplus < 201703L
     OutputFragments register output_frags;
+#else
+    OutputFragments output_frags;
+#endif
     int32_t key_start = 0;
     int32_t key_end = p.num_keys / kBlockSizeJ * kBlockSizeJ;
     for (; key_start < key_end; key_start += kBlockSizeJ) {