changed parametric type error in normalize to a warning (#57183)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/57183
Previously, if it was unable to support matching against a type, it would throw an error.
However, this exposes the user to arbitrary Torchscript schemas, which may or may not be problematic. Although we may support these in the future, for now we just return False (which will simply eliminate that schema from the candidates).
Test Plan: T89661626 and T89664016
Reviewed By: spaugh, khabinov
Differential Revision: D28072018
fbshipit-source-id: 83017d1e96d19912163edc74a5e43b2816783218
