Add operators to Partitioner (#3407)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/3407

1. Register aten operators in the vulkan partitioner.
2. Fix some minor operators name issue due to mismatch between the torch api and actual aten name

Note: Permute is not yet registered due to tensor movement issues with the "Partial" model where the `Linear` operator is decomposed into `permute` and `addmm`. Will fix in later diffs.
ghstack-source-id: 224398849

Reviewed By: SS-JIA

Differential Revision: D56695929

fbshipit-source-id: 80cd684caa8d15179b4514945f58e67d2893797f