CRDT sync handles parts of speech

[rmunn]

Fix #1178.

Parts of speech now sync both ways in FW and CRDT sync. Someday we may want to generalize this to include more types of CmPossibilityLists but for now parts of speech are working.

Semantic domains will be handled in a separate PR.

[rmunn]

Several bits of this code I would like to discuss with @hahn-kev before merging the PR. But the new unit tests are passing, and the code looks generally all right to me.

[hahn-kev]

Nice work, just some minor changes, left a comment to start a discussion about if we want to even keep around the in memory implementation.

[rmunn]

@hahn-kev - All requested changes are done.

Only thing I'm not sure of is the CreatePartOfSpeech(input) implementation for the dry run API. I have it just returning the input and not trying to call api.GetPartOfSpeech(input.Id) since that's almost certain to return null. However, I don't know if that's always going to be the case: I could see there being a situation, in certain tests perhaps, where CreatePartOfSpeech is being called in a situation where the api already has a PoS with that GUID.

So perhaps I should have it return api.GetPartOfSpeech(input.Id) ?? input instead. Thoughts?

[hahn-kev]

I'm fine with returning the input as that's what we do in the other APIs

[rmunn]

@hahn-kev - I've rebased on top of develop and fixed the merge conflicts. I did an interactive rebase so that I could edit the commits that added the PartOfSpeechSync code, and I rewrote them to use their own PartOfSpeechSync.cs static class similar to EntrySync etc.

Tests are still passing. Everything should be ready to re-review and merge.