refactor: separate aggregate group completion from input ordering

[xavlee]

Which issue does this PR relate to?

• Part of Supporting analytics over large amounts of pre‑partitioned data #24438.
• Builds on test: cover unsorted contiguous groups in one partition #24737.

Rationale for this change

Aggregate execution currently derives both input-order metadata and the runtime mechanism for recognizing completed groups from InputOrderMode.

Completing a group incrementally is contingent on knowing that "after processing a group, we will never see this group again in this stream". As such, input ordering covers only a subset of valid group completion cases. A sorted input guarantees that after processing a group, we will never see it again; however, the general case proves that so long as all rows in a groups are contiguous, the group completion property is satisfied regardless of the input ordering.

Consider the following:

AAABBBCCCC -> sorted, can be emitted incrementally
CCCAAAABBB -> unsorted, can be emitted incrementally still 

This PR introduces a new GroupCompletionMode that decouples the input ordering from group completion semantics. Currently, we directly map existing InputOrderMode to this property. In a follow up PR, we will read group contiguity expressions to evaluate this property.

What changes are included in this PR?

• Introduce private GroupCompletionMode::{None, Partial, Full} variants.
• Derive GroupCompletionMode from InputOrderMode during AggregateExec construction.
• Pass GroupCompletionMode through ordered aggregate tables, streams, and spill replay paths.
• Use InputOrderMode for required ordering, output ordering, and maintains_input_order metadata.
• Retain GroupOrdering::try_new(&InputOrderMode) as the public compatibility entry point.

The ordering-derived conversion is:

InputOrderMode::Linear                    -> GroupCompletionMode::None
InputOrderMode::PartiallySorted(indices) -> GroupCompletionMode::Partial(indices)
InputOrderMode::Sorted                    -> GroupCompletionMode::Full

Stack

1. #24737 — test: cover unsorted contiguous groups in one partition
2. #24697 — refactor: separate aggregate group completion from input ordering ← this PR
3. #24698 — feat: add narrow group-contiguous source property
4. #24497 — feat: stream exact group-contiguous aggregates

Are these changes tested?

The characterization from #24737 remains Linear, None, Final, and single-hash after the refactor.

Are there any user-facing changes?

No. GroupCompletionMode is crate-private, and the public GroupOrdering constructor is unchanged.

Review this layer

View only this PR layer

[codecov-commenter]

Codecov Report

❌ Patch coverage is 85.41667% with 14 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.42%. Comparing base (ee59f62) to head (372429f).

Files with missing lines	Patch %	Lines
datafusion/physical-plan/src/aggregates/mod.rs	82.00%	2 Missing and 7 partials ⚠️
...tafusion/physical-plan/src/aggregates/order/mod.rs	83.33%	2 Missing ⚠️
.../aggregates/aggregate_hash_table/common_ordered.rs	50.00%	0 Missing and 1 partial ⚠️
...hysical-plan/src/aggregates/grouped_hash_stream.rs	0.00%	0 Missing and 1 partial ⚠️
...ysical-plan/src/aggregates/ordered_final_stream.rs	92.85%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #24697      +/-   ##
==========================================
- Coverage   81.42%   81.42%   -0.01%     
==========================================
  Files        1121     1121              
  Lines      402142   402197      +55     
  Branches   402142   402197      +55     
==========================================
+ Hits       327460   327501      +41     
- Misses      55484    55490       +6     
- Partials    19198    19206       +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.