feat(non linear font scaling): optimization: cache any generated tables to...

feat(non linear font scaling): optimization: cache any generated tables to reduce duplicate calculations

Some clients (e.g. Compose) might call this every time they convert, in
which case we need to ensure it is fast and avoids unnecessary
allocations.

Bug: b/239736383
Flag: android.content.res.font_scale_converter_public
Test:  atest FrameworksCoreTests:android.content.res.FontScaleConverterFactoryTest

Change-Id: Ie73fa6d49e7605eea12a37930c82db7747dd8e2c