ARM/ARM64: Improve Mterp */lit8 assembly.

Embed shifts in operations when possible. Use UBFX for shift
distance extraction - for ARM this saves an instruction, for
ARM64 this just improves readability as the reader doesn't
need to know that the higher bits are unused by shifts.

Test: Run ART test suite on Nexus 9.
Change-Id: I5666f3eaded618d3f6cb754c4a431fd9548e746d