use edge_compile_config in verifier (#3521)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/3521

this diff use `edge_compile_config` to create edge verifier, to replace the original `enable` and `_check_edge_ops`. This update has three advantages:

1. User friendly. Both `enable` and `_check_edge_ops` are from `edge_compile_config`, in the previous setting user needs to manually extract the two attributes from their compile_config then forward them to verifier. Now user can just forward their config to verifier, without extra work.
2. Consistency. In ET, we use `EdgeCompileConfig` to control how we lower compile model graph to edge graph. We should use the same config to test whether or not the graph is as we expected, rather than manually set the test criteria.
3. Decreaase reduntancy. Directly using `EdgeCompileConfig` can decrease redundancy, expecially if new attribute has been introduced to `EdgeCompileConfig`, or old need to be removed. Otherwise we need to take forever to add/remove options to verifier

Reviewed By: larryliu0820

Differential Revision: D57010390

fbshipit-source-id: 10e550db997cf4fa039b8f6de150cfb04f700d7a