[PyTorch Edge] Reduce the cost of computing isIncludedInAlias() (#65735)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/65735
Currently, `isIncludedInAlias()` calls `getRuntimeDispatchKeySet()` which creates a new `DispatchKeySet` object from an enumerated list of dispatch keys. `isIncludedInAlias()` then checks if a single dispatch key is part of this set. Instead, just pass in the key one wishes to check. This is marginally faster.
ghstack-source-id: 139281528
Test Plan:
See these 2 AI Bench Runs on the Milan-FFF-11-30 device.
### Before
[AI Bench](https://www.internalfb.com/intern/aibench/details/237302972704466), [Flamegraph](https://interncache-all.fbcdn.net/manifold/aibench/tree/mobile/pt/profiling_reports/speech_transducer_v25_perf_1632804218329.html)
### After
[AI Bench](https://www.internalfb.com/intern/aibench/details/606320012968375), [Flamegraph](https://interncache-all.fbcdn.net/manifold/aibench/tree/mobile/pt/profiling_reports/speech_transducer_v25_perf_1632807348803.html)
Check the the flamegraphs, and focus on any kernel registration code path during library initialization.
Reviewed By: swolchok
Differential Revision: D31228062
fbshipit-source-id: 7a986e3593c30ded7919cd3b564ec579dc97ab5f
