fix: preserve null_aware on logical JoinNode proto round-trip

[mithuncy]

Summary

Closes #22065.

null_aware was missing from JoinNode in the logical proto
(it was added to the physical HashJoinExecNode in #19635). The
encoder dropped it via .. destructuring and the decoder had no
field to restore it from, so any to_proto -> from_proto round
trip silently downgraded a null-aware LeftAnti (NOT IN semantics)
to a plain LeftAnti and returned wrong rows.

Changes

• Proto: add bool null_aware = 9; to JoinNode (backward compatible).
• Encoder/decoder plumb null_aware through; decoder switches to
  Join::try_new so the field is set at construction (also fixes
  null_equality, same bug on the same path).
• Regression test for NOT-IN-with-NULL round trip.

Test plan

• cargo test -p datafusion-proto --test proto_integration cases::roundtrip_logical_plan passes (incl. new roundtrip_join_null_aware).
• Clippy clean.

[github-actions]

Thank you for opening this pull request!

Reviewer note: cargo-semver-checks reported the current version number is not SemVer-compatible with the changes in this pull request (compared against the base branch).

Details

     Cloning apache/main
    Building datafusion-proto v53.1.0 (current)
       Built [  57.204s] (current)
     Parsing datafusion-proto v53.1.0 (current)
      Parsed [   0.144s] (current)
    Building datafusion-proto v53.1.0 (baseline)
       Built [  56.795s] (baseline)
     Parsing datafusion-proto v53.1.0 (baseline)
      Parsed [   0.141s] (baseline)
    Checking datafusion-proto v53.1.0 -> v53.1.0 (no change; assume patch)
     Checked [   2.391s] 222 checks: 221 pass, 1 fail, 0 warn, 30 skip

--- failure constructible_struct_adds_field: externally-constructible struct adds field ---

Description:
A pub struct constructible with a struct literal has a new pub field. Existing struct literals must be updated to include the new field.
        ref: https://doc.rust-lang.org/reference/expressions/struct-expr.html
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.47.0/src/lints/constructible_struct_adds_field.ron

Failed in:
  field JoinNode.null_aware in /home/runner/work/datafusion/datafusion/datafusion/proto/src/generated/prost.rs:392
  field JoinNode.null_aware in /home/runner/work/datafusion/datafusion/datafusion/proto/src/generated/prost.rs:392

     Summary semver requires new major version: 1 major and 0 minor checks failed
    Finished [ 119.173s] datafusion-proto

[neilconway]

Overall this is a well-done PR! Nice work, @mithuncy

It does make me a bit concerned that an issue like this surfaced. It would be great if we had a systematic way to test that query plans roundtrip through proto, like we do for substrait (datafusion/sqllogictest/src/engines/datafusion_substrait_roundtrip_engine). That could be worth taking on as a follow-up.

[neilconway]

Rather than first constructing the join with an incorrect null_aware field and then overwriting it here, can we try ensuring that the null_aware field is correct when the join is constructed? That would be more robust and avoids the intermediate period where we've constructed a value but it is in an incorrect state.

i.e., something like https://gist.github.com/neilconway/2dc2449dcf4cca3a60e1861e445e042c

[mithuncy]

@neilconway Thanks a lot — much cleaner than mine, and it also fixes the null_equality drop on the same path.
Edited the now-obsolete null_equality related paragraph from the PR description. and LGTM