Merge pull request #4460 from RosettaCommons/vmullig/multithreaded_scoring_incremental_merge2
Incremental merge 2 for multithreaded scoring
This PR relaxes some of the locking of the global options system. I was overdoing it, and it was hurting multithreaded performance and performance scaling a bit.
(Note that the full locking can be restored by defining `OPTIONS_COLLECTION_FULL_THREAD_SAFETY`. This is not defined by default. When `MULTI_THREADED` is defined, as it is when building with `extras=cxx11thread`, only the necessary locking occurs.)
- [x] Pull request #4459 must be merged before this one.
@bcov77