Implement Tanh Gelu Approximation (#61439)

Summary:
1. Implements pytorch/pytorch#39853
2. Adds approximate boolean flag to Gelu
3. Enables Tanh Gelu approximation
4. Adds double backward support for Gelu
5. Enable Tanh Gelu in NvFuser

```
def gelu(x, approximate : str = 'none'):
    if approximate == 'tanh':
        # sqrt(2/pi) = 0.7978845608028654
        return 0.5 * x * (1.0 + torch.tanh(0.7978845608028654 * (x + 0.044715 * torch.pow(x, 3.0))))
    else:
        return x * normcdf(x)
```

Linking XLA PR - pytorch/xla#3039

Pull Request resolved: pytorch/pytorch#61439

Reviewed By: mikaylagawarecki

Differential Revision: D33744717

Pulled By: jbschlosser

fbshipit-source-id: d64532a562ed53247bb4fa52bb16722634d5c187

nestedtensor/csrc/activation.cpp

@@ -8,15 +8,15 @@ namespace F = torch::nn::functional;
 
 namespace at {
 
-Tensor NestedTensor_gelu(const Tensor& self) {
+Tensor NestedTensor_gelu(const Tensor& self, const int64_t approximate) {
   if (is_nested_tensor_impl(self) && get_is_contiguous(self)) {
     return wrap_buffer(
         at::gelu(get_buffer(self)),
         get_efficient_nested_size(self),
         get_efficient_nested_stride(self));
   }
   return map_nested_tensor(
-      [](at::Tensor tensor) { return at::gelu(tensor); }, self);
+      [&approximate](at::Tensor tensor) { return at::gelu(tensor, approximate); }, self);
 }
 
 Tensor NestedTensor_elu(const Tensor& self, const Scalar& alpha, const Scalar& scale, const Scalar& input_scale) {