refactor(proto): migrate aggregate exec serde

[Phoenix500526]

Which issue does this PR close?

• Closes Proto: migrate AggregateExec (function-codec path) #23512.

Rationale for this change

Aggregate protobuf conversion lived in the central proto crate, which prevented AggregateExec from owning its function-codec serialization.

Move encoding and decoding into AggregateExec and keep the deprecated helpers as delegates so existing callers retain wire-compatible behavior.

What changes are included in this PR?

Are these changes tested?

Yes

Are there any user-facing changes?

[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-physical-plan v54.1.0 (current)
       Built [  42.913s] (current)
     Parsing datafusion-physical-plan v54.1.0 (current)
      Parsed [   0.145s] (current)
    Building datafusion-physical-plan v54.1.0 (baseline)
       Built [  38.094s] (baseline)
     Parsing datafusion-physical-plan v54.1.0 (baseline)
      Parsed [   0.141s] (baseline)
    Checking datafusion-physical-plan v54.1.0 -> v54.1.0 (no change; assume patch)
     Checked [   0.674s] 223 checks: 223 pass, 30 skip
     Summary no semver update required
    Finished [  83.510s] datafusion-physical-plan
    Building datafusion-proto v54.1.0 (current)
       Built [  61.748s] (current)
     Parsing datafusion-proto v54.1.0 (current)
      Parsed [   0.019s] (current)
    Building datafusion-proto v54.1.0 (baseline)
       Built [  61.049s] (baseline)
     Parsing datafusion-proto v54.1.0 (baseline)
      Parsed [   0.020s] (baseline)
    Checking datafusion-proto v54.1.0 -> v54.1.0 (no change; assume patch)
     Checked [   0.241s] 223 checks: 222 pass, 1 fail, 0 warn, 30 skip

--- failure trait_method_marked_deprecated: trait method #[deprecated] added ---

Description:
A trait method is now #[deprecated]. Downstream crates will get a compiler warning when using this method.
        ref: https://doc.rust-lang.org/reference/attributes/diagnostics.html#the-deprecated-attribute
       impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.49.0/src/lints/trait_method_marked_deprecated.ron

Failed in:
  method try_into_aggregate_physical_plan in trait datafusion_proto::physical_plan::PhysicalPlanNodeExt in /home/runner/work/datafusion/datafusion/datafusion/proto/src/physical_plan/mod.rs:664
  method try_from_aggregate_exec in trait datafusion_proto::physical_plan::PhysicalPlanNodeExt in /home/runner/work/datafusion/datafusion/datafusion/proto/src/physical_plan/mod.rs:664

     Summary semver requires new minor version: 0 major and 1 minor checks failed
    Finished [ 124.295s] datafusion-proto

[codecov-commenter]

Codecov Report

❌ Patch coverage is 77.10438% with 68 lines in your changes missing coverage. Please review.
✅ Project coverage is 80.63%. Comparing base (4a40101) to head (0bb42c4).

Files with missing lines	Patch %	Lines
datafusion/physical-plan/src/aggregates/mod.rs	80.85%	23 Missing and 31 partials ⚠️
datafusion/proto/src/physical_plan/mod.rs	6.66%	14 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #23779      +/-   ##
==========================================
- Coverage   80.65%   80.63%   -0.03%     
==========================================
  Files        1091     1091              
  Lines      370692   370720      +28     
  Branches   370692   370720      +28     
==========================================
- Hits       298972   298916      -56     
- Misses      53885    53961      +76     
- Partials    17835    17843       +8     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[adriangb]

Please resolve conflicts 🙏🏻

[adriangb]

Reviewed this against the reference implementation in #23496 and against the code it deletes
from datafusion/proto/src/physical_plan/mod.rs — the port looks faithful field-for-field
(encode/decode struct construction, by-name AggregateMode match, asc: !descending ordering
encoding, (!buf.is_empty()).then_some(buf) UDAF payload semantics, and the payload → codec /
registry → codec lookup order all match). Two things to fix up before merge:

1. Please move the unit tests along with the function. This PR moves
   split_human_display_alias from datafusion-proto into aggregates/mod.rs, but the two
   tests covering it — split_human_display_alias_ignores_mismatched_alias and
   split_human_display_alias_keeps_malformed_prefix_literal — are deleted rather than moved.
   They cover exactly the tricky cases in that helper (mismatched alias, malformed prefix
   falling back to the literal), so they should follow it into datafusion-physical-plan.

2. Consider keeping the explanatory comments from [EPIC/reference] Port ExecutionPlan serialization to try_to_proto / try_from_proto hooks (full) #23496. The port renames locals and strips
   the reference version's comments. The renames are fine, but a couple of the comments are
   load-bearing rationale rather than narration — in particular the note that AggregateMode is
   matched by name because the proto and datafusion_common enums are numbered differently,
   so a numeric cast would silently corrupt the wire format. That is the kind of thing a future
   editor needs to see at the call site. Same for the note that encode_udaf/decode_udaf
   already apply the (!buf.is_empty()).then_some(buf) semantics, and that the aggregate's own
   physical schema (not the child schema) is what arg/ordering/filter exprs decode against.

Also note HUMAN_DISPLAY_ALIAS_PREFIX / encode_human_display_alias now exist in both crates
(the deprecated serialize_physical_aggr_expr path still needs the datafusion-proto copy).
That's expected for now, but worth a comment on one of them pointing at the other so the two stay
in sync until the deprecation window closes.

[Phoenix500526]

Please resolve conflicts 🙏🏻

Done

[Phoenix500526]

Reviewed this against the reference implementation in #23496 and against the code it deletes from datafusion/proto/src/physical_plan/mod.rs — the port looks faithful field-for-field (encode/decode struct construction, by-name AggregateMode match, asc: !descending ordering encoding, (!buf.is_empty()).then_some(buf) UDAF payload semantics, and the payload → codec / registry → codec lookup order all match). Two things to fix up before merge:

1. Please move the unit tests along with the function. This PR moves
   split_human_display_alias from datafusion-proto into aggregates/mod.rs, but the two
   tests covering it — split_human_display_alias_ignores_mismatched_alias and
   split_human_display_alias_keeps_malformed_prefix_literal — are deleted rather than moved.
   They cover exactly the tricky cases in that helper (mismatched alias, malformed prefix
   falling back to the literal), so they should follow it into datafusion-physical-plan.
2. Consider keeping the explanatory comments from [EPIC/reference] Port ExecutionPlan serialization to try_to_proto / try_from_proto hooks (full) #23496. The port renames locals and strips
   the reference version's comments. The renames are fine, but a couple of the comments are
   load-bearing rationale rather than narration — in particular the note that AggregateMode is
   matched by name because the proto and datafusion_common enums are numbered differently,
   so a numeric cast would silently corrupt the wire format. That is the kind of thing a future
   editor needs to see at the call site. Same for the note that encode_udaf/decode_udaf
   already apply the (!buf.is_empty()).then_some(buf) semantics, and that the aggregate's own
   physical schema (not the child schema) is what arg/ordering/filter exprs decode against.

Also note HUMAN_DISPLAY_ALIAS_PREFIX / encode_human_display_alias now exist in both crates (the deprecated serialize_physical_aggr_expr path still needs the datafusion-proto copy). That's expected for now, but worth a comment on one of them pointing at the other so the two stay in sync until the deprecation window closes.

Done