feat: add subquery representation

[cpcloud]

This PR adds support for correlated subqueries to substrait.

Notes:

1. It'll be easier to review commit by commit, since I had to combine
   expressions and relations together to get the protos to compile.
2. While there are actually only three kinds of subqueries (scalar, EXISTS and ANY),
   we chose to keep a few of the higher level representations so that information about
   the original intention is preserved.

Closes #119.

cc @pdet @Mytherin, would love your thoughts here!

[jacques-n]

Generally looks good. Interested to see if others agree.

[jacques-n]

This looks good to me but I'd like to get some more eyes on it.

[jacques-n]

I lied... we need to also add the docs part of this.

Once people agree with the shape of this, let's try to update the appropriate parts of the expressions/etc sections.

[Mytherin]

Looks good to me as a way of representing unflattened subqueries. The only potential missing feature I see here is support for the ARRAYsubquery, e.g. in Postgres:

select array(select * from generate_series(0,5,1));
--  {0,1,2,3,4,5}

[cpcloud]

Looks good to me as a way of representing unflattened subqueries. The only potential missing feature I see here is support for the ARRAYsubquery, e.g. in Postgres:

select array(select * from generate_series(0,5,1));
--  {0,1,2,3,4,5}

Thanks!

Yeah, functions that accept subqueries (or sets of rows) would be an interesting follow up.

This would likely also need work in the type system.

[cpcloud]

I will add some documentation to the spec later today about subqueries.

[jacques-n]

Thanks for working through this!