[fix](variant) Skip full footer scan when constructing VariantStatsCaculator

[csun5285]

Problem

In vertical compaction, SegmentWriter::init() is called multiple times against the same writer (key columns first, then each value-column group). The footer accumulates across calls, so every additional init() re-scans an ever-larger footer — including entries from prior init()s that the current column_ids cannot address.

	Before	After
Single construction	O(footer accumulated size)	O(this init's column_ids size)
N vertical-compaction init() calls	O(N²)	O(N)

Fix

Snapshot _footer.columns_size() before _create_writers appends new entries and pass it to VariantStatsCaculator as footer_column_offset. The constructor only walks [offset, end) — its own slice.

When performing compaction with 10,000 columns in the variant, this method accounts for around 8% of CPU usage. After changing it to O(N), this section has disappeared from the flame graph.

What problem does this PR solve?

Issue Number: close #xxx

Related PR: #xxx

Problem Summary:

Release note

None

Check List (For Author)

• Test

  • Regression test
  • Unit Test
  • Manual test (add detailed scripts or steps below)
  • No need to test or manual test. Explain why:
    • This is a refactor/code format and no logic has been changed.
    • Previous test can cover this change.
    • No code files have been changed.
    • Other reason

• Behavior changed:

  • No.
  • Yes.

• Does this need documentation?

  • No.
  • Yes.

Check List (For Reviewer who merge this PR)

• Confirm the release note
• Confirm test cases
• Confirm document
• Add branch pick label

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[csun5285]

run buildall

[hello-stephen]

BE Regression && UT Coverage Report

Increment line coverage 100.00% (4/4) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.43% (26782/37494)
Line Coverage	53.79% (279652/519859)
Region Coverage	47.07% (214603/455963)
Branch Coverage	50.43% (97260/192877)

[csun5285]

/review

[github-actions]

OpenCode automated review failed and did not complete.

Error: Review step was failure (possibly timeout or cancelled)
Workflow run: https://github.com/apache/doris/actions/runs/24952118629

Please inspect the workflow logs and rerun the review after the underlying issue is resolved.

[eldenmoon]

/review

[github-actions]

I did not find blocking issues in this PR.

Critical checkpoints:

• Goal / correctness: The change is narrowly targeted at the repeated SegmentWriter::init() reuse path in vertical compaction / segcompaction. Scoping VariantStatsCaculator to the footer slice added by the current init() matches how SegmentWriter accumulates footer entries across groups, so the fix addresses the reported stale-scan behavior without changing the downstream stats update logic.
• Scope: The modification is small and focused: one caller-side offset snapshot, one constructor parameter, and one unit test.
• Concurrency: No new concurrency risk is introduced here. The touched logic runs on the per-writer compaction path and the SegmentWriter reuse is serialized by the surrounding compaction flow.
• Lifecycle: The review-sensitive lifecycle detail is the persistent _footer across repeated clear() / init() cycles; this patch handles that explicitly by snapshotting the footer size before the current group appends its own metas.
• Config / compatibility: No config, protocol, or storage-format compatibility changes.
• Parallel paths: The fix covers both known repeated-init paths that reuse SegmentWriter (VerticalBetaRowsetWriter and segcompaction), because both go through SegmentWriter::init(col_ids, is_key).
• Special conditions: The new footer_column_offset condition is justified and documented in code.
• Test coverage: The added unit test captures the intended footer-offset behavior and checks that stats land on the current slice instead of stale entries from a prior init. I could not run the BE UT in this runner because thirdparty/installed/bin/protoc is missing, so my validation here is code inspection only.
• Observability: No additional observability appears necessary for this localized fix.
• Data / persistence: No transaction, persistence, or metadata replay semantics are changed beyond which in-memory footer entries are indexed during compaction.
• Performance: The optimization claim is consistent with the implementation; the constructor scan is reduced from accumulated footer size to the current init slice, which removes the repeated full-footer walk.
• Other issues: None found in the reviewed diff.

User focus points: No additional user-provided focus points were supplied for this review.

[eldenmoon]

LGTM

[github-actions]

PR approved by at least one committer and no changes requested.

[github-actions]

PR approved by anyone and no changes requested.