GH-774: Consoliate BitVectorHelper.getValidityBufferSize and BaseValueVector.getValidityBufferSizeFromCount

[rtadepalli]

What's Changed

Just removing some duplicate functions in anticipation of cleaning out some transferPair duplication across complex vectors.

Closes #774

[lidavidm]

Hmm. This is a protected API, so this is technically a breaking change. Albeit not one likely to be noticed. But I think we should mark it as such all the same, and perhaps consider making the next version 19.0.0 then. (Any thoughts from other reviewers?) The API has also been there for quite some time, though I suppose it's fine to clean things up when we notice redundancy.

[rtadepalli]

Sorry I think I am missing something here - nothing public is changing so I don't follow how this is a breaking change. Could you please clarify?

[lidavidm]

A protected method is still visible if a user defines their own subclass.

[rtadepalli]

Hmm. I guess indeed nothing is stopping people from defining their own custom vector types...

[lidavidm]

It's admittedly something I don't expect people to do, so I could be convinced to not consider this a breaking change.

[rtadepalli]

I guess nothing is being "removed" from the code base. If you / other reviewers have seen anything either on the mailing list or via GH issues about people implementing custom types then maybe marking this a breaking change is the way to go. If not then IMHO the change is low-key and likely won't impact anything. In the off chance someone is using custom types and they are impacted because of this, maybe we could do a minor release instead of a patch (don't know if that is any better).

[rtadepalli]

I've re-added BaseValueVector.getValidityBufferSizeFromCount back and have marked it deprecated. I personally care more about the removal of the duplicate than the method - I suppose it can be removed in a later release without forcing a breaking change right now.

cc @lidavidm