[TensorExpr] Add `TensorExprKernel::runFast` method. (#57328)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/57328
This method uses `CodeGen::call_raw` instead of `CodeGen::call`.
Test Plan: Imported from OSS
Reviewed By: bertmaher
Differential Revision: D28110359
Pulled By: ZolotukhinM
fbshipit-source-id: 4fdffc8196d24fc3300a9b4bc69f67562042a045
Author
Mikhail Zolotukhin
