[FEA] pyspark integration tests basic aggregate tests

[revans2]

Is your feature request related to a problem? Please describe.
We want a comprehensive set of tests for various aggregation operations across all of the supported types.

count - which we need to try both with null filtering and without, and with distinct and without
min, max - which should be bit for bit the same
sum, average - which has some leeway on floating point operations, but not for the others
first, last - which might be a little difficult because they are not guaranteed to return the same value, because of how shuffle works. We might need to have some specially crafted data sets that we can use to check for consistent answers. or operations to make this test work well in any condition, like pre-partition the data on the aggregation key.

We should also test these without a group-by so we get a single result for the entire data set.

[kuhushukla]

Resolved via commit sha = 10922c7