Parallel decompiler/disassembler performance

[sad-dev]

This p/r has 3 commits which do the following:

1.) Remove synchronized bottlenecks in ContextCache and SleighLanguage 2.) store HighFunction.func.getEntryPoint() and HighFunction.func.getEntryPoint().getAddressSpace() to avoid expensive hits to the database lock.

[sad-dev]

#6649 is an issue solved by the first commit

#2791 has some old discussion on bottlenecks in parallel decompilation that the changes here mitigate (but not completely remove)