[Static Runtime] Adding out-variant support for quantized::linear_relu_dynamic_fp16 (#73238)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/73238
ghstack-source-id: 149706142
Test Plan:
Tested on video model.
Profile before:
```
4.36922 ms. 17.1047%. quantized::linear_relu_dynamic_fp16 (14 nodes)
```
Profile after:
```
3.80852 ms. 17.1074%. quantized::linear_relu_dynamic_fp16 (14 nodes, out variant)
```
Reviewed By: mikeiovine
Differential Revision: D34287961
fbshipit-source-id: b88e2f3432215eac14fd36f945a4810d29ba1051
(cherry picked from commit 076a766ab471c362af2f2ee3b55fe75829f5e955)