Support FlatHashStrategy with 100_000+ columns

[dain]

Release notes

( ) This is not user-visible or is docs only, and no release notes are required.
( ) Release notes are required. Please propose a release note for me.
( ) Release notes are required, with the following suggested text:

# Section
* Fix some things. ({issue}`issuenumber`)

[sopel39]

approach looks ok. I don't think it can impact perf, can it?

[sopel39]

test failures seem relevant

[sopel39]

It probably should keep track of chunk first channel index, e.g: 0, 500, 1000, so that correct offsets are used when accessing: byte[] fixedChunk, byte[] variableChunk and BlockBuilder[] blockBuilders);

[dain]

I assume there is a bug like this somewhere in here, but my memory is that all of the indexes come from KeyField and not the oridinal of the key field object in the list… If you look at the code, I don’t think we are using an indexed for loop anywhere.

[dain]

I don't think it can impact perf, can it?
I doubt it. The method just call though to one or more statics. It is possible that smaller methods help as large methods cause vm optimizers to be disabled. I picked 500 arbitrarily and it is possible 100 is better for this reason