Remove optimized_for_inference (#1012)
Summary:
Now we are enabling ofi for jit inference by default (https://github.com/pytorch/benchmark/blob/main/torchbenchmark/util/backends/jit.py#L4), so remove optimized_for_inference in metadata.
Pull Request resolved: https://github.com/pytorch/benchmark/pull/1012
Reviewed By: aaronenyeshi
Differential Revision: D37702454
Pulled By: xuzhao9
fbshipit-source-id: 92b347ff50a8be28cd4d4e65e58a360540a6c428
