NodeId and NodePath are excess in NodeVersion fetching methods

[vbradnitski]

To refactor and deprecate.

Methods like:

NodeService.getByIdAndVersionId NodeStorageService.get NodeStorageService.getNode NodeStorageService.getVersion

Classes like: GetNodeByIdAndVersionIdCommand

[rymsha]

Reasons: After work on #9205 it became evident that a few methods with NodeVersion parameter also require NodeId that is only used for ES routing. Routing is worthless when there is only one shard per index - like in our case. (exception is parent/child index, when parent is not known, but we can use ANY routing, it will still land on the same single shard)

• Sometimes NodeId is not known and methods that need to call such method do a worthless query trying to guess correct NodeId.
• NodeStorageService.get and NodeStorageService.getNode are very similar, but get also requests branchService, to construct a bogus node with "previous" data but some fields overridden with "current" values. It is confusing, especially because method is called "get" and may be used incorrectly.

[rymsha]

NodeService.getByIdAndVersionId should still require NodeId (on high level we need to be sure NodeVersion has correct NodeId) The rest of the methods will be cleaned up in #9930