Unify simplification passes after high-level optimizations
* Add a constant folding pass after the later high-level optimizations
* Remove dead_code_elimination$after_lse to unify simplification passes (always constant folding, instruction simplifier, dead code elimination)
* Make dead_code_elimination$before_codegen run always.
Locally speed compile Pixel 5:
* system server -64.28KB (-0.13%)
* SystemUIGoogle: -16.31KB (-0.06%)
* AGSA: -140.57 (-0.04%)
Test: art/test/testrunner/testrunner.py --host --64 --optimizing -b
Change-Id: I305877fd83471dff059fe1d4f5b98e5b1063f1a5
27 files changed
