Atomic updates for synchronised language resources

For now synchronisation works on Language Resource lvl and can be run for the whole dataset that actual LR provides.
For example right now we can only only recreate glossary in third party service using all the terms from Term Collection.

What is desired to be implemented (!!Unable to render embedded object: File (as an example) not found.!!)- is a way to update 3rd party glossary term when this term is updated in TermPortal, if that is possible with for 3rd party API of course. 

So we need to implement a way to make atomic updates.
Sure thing this should be done with abstraction layer with decoupled contexts: in a given moment of a time we can't guess what 3rd party services may come and how to work with them, and some LR types come with different Private or Public plugins and may simply be not present in clients environment.