Properly handle array key cols in query operations

[rbasralian]

In something like the below, the equivalent arrays should be treated as equal and aggregated/joined accordingly. (It already seems to work for vectors.)

arr1 = new String[] { 'abc', 'def'}
arr2 = new String[] { 'abc', 'def'}
myTable_arr = emptyTable(2).update("KeyCol = i==0 ? arr1 : arr2", "X = ii + 1")
mySum_arr = myTable_arr.sumBy("KeyCol")
myJoin_arr = myTable_arr.where("i==0").naturalJoin(myTable_arr.where("i==1"), "KeyCol", "X_joined=X")


vec1 = new io.deephaven.vector.ObjectVectorDirect(arr1)
vec2 = new io.deephaven.vector.ObjectVectorDirect(arr2)
myTable_vec = emptyTable(2).update("KeyCol = i==0 ? vec1 : vec2", "X = ii + 1")
mySum_vec = myTable_vec.sumBy("KeyCol")
myJoin_vec = myTable_vec.where("i==0").naturalJoin(myTable_vec.where("i==1"), "KeyCol", "X_joined=X")

(https://deephaven.atlassian.net/browse/DH-11038 in Legacy engine)

[devinrsmith]

I think this is due to io.deephaven.chunk.util.hashing.ObjectChunkEquals not using deepEquals. As part of a PR that will merge soon, #5225, I've added a new version that uses deepEquals. We'll still need to plumb it properly to get it to work here.

[devinrsmith]

It looks like DHE has moved to a per array-type solution to this (as opposed to the more generic deepEquals). This seems like a good solution for DHC as well, as deepEquals is potentially much less efficient.

[devinrsmith]

https://github.com/deephaven-ent/iris/commit/1d10bf7e8fe460a3e3047178a60dabd06fbc7588