Support predicate-pushdown for isin operations

[rjzamora]

Closes #607

This PR adds predicate-pushdown support for dd._Frame.isin(values), and uses a somewhat-ugly workaround to account for the fact that dask.dataframe will produce a distinct MaterializedLayer object for the values argument to isin.

Note that this PR also addresses "not in" support. However, dask.dataframe will produce the wrong result for these filters until dask/dask#10320 is merged.

[codecov-commenter]

Codecov Report

Merging #1160 (5193718) into main (00aea43) will increase coverage by 0.22%.
The diff coverage is 83.78%.

❗ Your organization is not using the GitHub App Integration. As a result you may experience degraded service beginning May 15th. Please install the Github App Integration for your organization. Read more.

@@            Coverage Diff             @@
##             main    #1160      +/-   ##
==========================================
+ Coverage   81.30%   81.53%   +0.22%     
==========================================
  Files          78       78              
  Lines        4414     4474      +60     
  Branches      799      817      +18     
==========================================
+ Hits         3589     3648      +59     
+ Misses        649      641       -8     
- Partials      176      185       +9     

Impacted Files	Coverage Δ
dask_sql/physical/utils/filter.py	80.88% <83.33%> (+3.04%)	⬆️
dask_sql/_compat.py	100.00% <100.00%> (ø)

... and 1 file with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[rjzamora]

@ayushdg @randerzander - Note that I added isna support to this PR, but I'd be happy to split that change into a separate PR if you prefer.

[ayushdg]

Changes generally lgtm!
Doesn't need to be in this pr, but as a followup it might also make sense to add simpler unit tests that just use the attempt predicate pushdown method on dask dataframes and add them to https://github.com/dask-contrib/dask-sql/tree/main/tests/unit

[rjzamora]

@ayushdg - I'm thinking we should layer on something like #1140 as a follow up. Let me know what you think.

[ayushdg]

Yeah it makes sense to add something that conceptually achieves what #1140 does without some of the extra logic/handling in that pr. We can address that in a followup and some of the logic from #1140 should be redundant with dask-cudf 22.06 and newer

[ayushdg]

Thanks a lot for looking into this @rjzamora!