Revert "Add zip() for vectorizing matrix lookups for paired indices"

[WardBrian]

Reverts #3411

I think this was a misreading of the request in stan-dev/stanc3#1704.

The zip function described in that issue would construct an array of pairs, as is common in every other language. The code to then index a matrix with an array of pairs should live in stan-dev/stan not stan-dev/math.

[SteveBronder]

Talking to Bob this was the signature we wanted. The zip as written in the doc would need a to_vector

[WardBrian]

That would depend on how the indexing operation was written in stan, no?

At the very least, this function should not be called zip, since that has a pretty well established meaning elsewhere. The original issue used the name zip_index, which would be fine

[SteveBronder]

If we returned a tuple of indices I just don't see how we would write something that would be more optimal than just writing the for loop. I think calling it flatten or zip_index would make sense

[SteveBronder]

Making the new pr rn closing this