Improve Bitmap32 Common Vector Rendering: faster + more accurate on all targets

* https://github.com/servo/pathfinder
* https://skia.org/docs/dev/design/aaa/
* https://docs.google.com/presentation/d/16r9HMS4_UBrcF3HUHscAqbSgkrtIwqaihZNwGP2TL_s/edit#slide=id.p
* https://docs.google.com/document/d/17Gq-huAf9q7wA4MRfXwpi_bYLrVeteKcSfAep0Am-wA/edit#

If we can have a vector rendering that works fast on all the targets, it would be awesome. So vector rendering would be consisent, and will work on all the targets automatically without specific GDI+, apple vector specific code.

Start by reviewing the linked Pathfinder and Skia references, along with the linked presentation and design document, then locate the Bitmap32 common vector-rendering entry points. The issue does not name files or tests. Done means vector rendering is faster, more accurate, consistent across targets, and no longer depends on target-specific GDI+ or Apple vector code.

Written by the indexing model from the issue text.