Will Spatial transformer operator be supported in the near future?

`#!/usr/bin/env python3

-- coding: utf-8 --

import torch
import torch.nn as nn
import torch.nn.functional as F

class STNet(nn.Module):
def init(self):
super(STNet, self).init()

    self.localization = nn.Sequential(
            nn.Conv2d(3, 32, kernel_size=3),
            nn.MaxPool2d(2, stride=2),
            nn.ReLU(True),
            nn.Conv2d(32, 32, kernel_size=5),
            nn.MaxPool2d(3, stride=3),
            nn.ReLU(True)
            )

    self.fc_loc = nn.Sequential(
            nn.Linear(32 * 14 * 2, 32),
            nn.ReLU(True),
            nn.Linear(32, 3 * 2)
            )

    self.fc_loc[2].weight.data.zero_()
    self.fc_loc[2].bias.data.copy_(torch.tensor([1, 0, 0, 0, 1, 0], dtype=torch.float))
    
def forward(self, x):
    xs = self.localization(x)
    xs = xs.view(-1, 32 * 14 * 2)
    theta = self.fc_loc(xs)
    theta = theta.view(-1, 2, 3)

    x = self.f32fwd(x, theta)

    return x

@torch.cuda.amp.custom_fwd(cast_inputs=torch.float32)  # TODO 在 pytorch 1.6.1 中移除: https://github.com/pytorch/pytorch/issues/42218
def f32fwd(self, x, theta):
    grid = F.affine_grid(theta, x.size(), align_corners=False)
    x = F.grid_sample(x, grid, align_corners=False)
    return x

`
The above is the Spatital transformer network, the "F.affine_grid" and "F.grid_sample" operators are not supported in the latest onnx opset version. I deeply hope STN will be supported in the next version as the importance of the transformer operator the in deep network. Besides, Can I export the STN to onnx and run inference by onnxruntime by other way in this opset version?
Thanks!

Start with the provided PyTorch STN example, focusing on F.affine_grid and F.grid_sample and their ONNX export behavior. Investigate the latest ONNX opset support and whether onnxruntime can execute an exported equivalent. Done would mean documented support for the operators or a verified alternative export and inference path.