[C++] Support SimplifyWithGuarantee for is_in expressions

[larry98]

Describe the enhancement requested

We'd like to use parquet predicate pushdown on is_in expressions, but this currently isn't supported in SimplifyWithGuarantee. We implemented a proof of concept where we sort and deduplicate the is_in expression's value set, then have SimplifyWithGuarantee binary search on the inequality bound and slice the value set accordingly. This works well, but I'm not sure what the correct interface for enabling this code path should be. Our current approach adds a new field to SetLookupOptions which allows the user to declare whether the value set is pre-sorted and deduplicated.

Any thoughts? I'd be happy to put up a PR if we agree on an interface.

Component(s)

C++

[mapleFU]

@larry98 feel free to do this and @ me if you finish

This requires some enhancement on KnownFieldValues in this function. Besides the syntax of "in" in SQL with null would be a bit hacking, which should be take care of

[larry98]

@mapleFU I opened a PR (#43256) with the change that I was thinking of. I don't fully follow the point about KnownFieldValues - I think I may be proposing something simpler.

[bkietz]

Issue resolved by pull request 43761
#43761