fix(query): avoid rewrites for Parquet-compatible decimal widening

[youngsofun]

I hereby agree to the terms of the CLA available at: https://docs.databend.com/dev/policies/cla/

Summary

ALTER TABLE ... MODIFY COLUMN currently rewrites the full table whenever a Decimal type changes, even when only the precision is widened and the persisted Parquet representation is unchanged. On large tables this turns a lossless schema change into a full scan and rewrite.

This PR makes eligible Decimal precision widening metadata-only when:

• scale is unchanged;
• precision only increases;
• the Decimal runtime kind is unchanged;
• the Parquet physical type is unchanged; and
• FIXED_LEN_BYTE_ARRAY uses the same byte width.

Changes that cross a Decimal kind or Parquet-width boundary, change scale, or narrow precision continue to rewrite data.

Schema Decimal variants are not stable across meta protobuf round trips, so the decision derives Decimal width from precision rather than enum discriminants. Persisted statistics also keep their original DecimalSize; schema-aware ColumnStatistics and ClusterStatistics views temporarily retag compatible bounds and fail open for incompatible statistics.

Reducers, range pruning, TopN, REPLACE, partition Top-K, runtime scan ordering/filtering, and reclustering now consume the safe statistics abstraction. Compatible Decimal TopN values merge at the widest precision, while incompatible statistics are discarded without blocking writes. Missing statistics remain unknown instead of being represented by fabricated bounds.

The persisted statistics layout is unchanged. Existing segment and snapshot metadata remain readable, and metadata-only DDL preserves existing block locations.

Tests

• Unit Test
• Logic Test
• Benchmark Test
• No Test - Explain why

Validated with:

• cargo test -p databend-storages-common-table-meta --lib
• cargo test -p databend-common-storages-fuse --lib cluster_statistics::tests
• cargo test -p databend-query --test it parquet_rs::prune_pages -- --nocapture
• cargo test -p databend-query --test it parquet_rs::prune_row_groups -- --nocapture
• cargo clippy -p databend-storages-common-table-meta -- -D warnings
• cargo fmt --all -- --check

The logic test covers metadata-only ALTER, predicate reads, post-ALTER writes, TopN, REPLACE, ANALYZE, COMPACT, reclustering, time travel, Decimal physical-width boundaries, and preservation of existing block locations.

Type of change

• Bug Fix (non-breaking change which fixes an issue)
• New Feature (non-breaking change which adds functionality)
• Breaking Change (fix or feature that could cause existing functionality not to work as expected)
• Documentation Update
• Refactoring
• Performance Improvement
• Other (please describe):

AI assistance

• AI usage: An AI coding agent assisted with root-cause analysis, storage and statistics implementation, regression tests, local review, and PR drafting
• Responsible human: @youngsofun
• The responsible human has read every line of this diff and can explain each change

---

This change is 

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: cdc2c24bde

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[github-actions]

🤖 CI Job Analysis

Workflow: 34074383775

⛔️ CANCELLED

Higher priority request detected - retry cancelled to avoid conflicts.

View Workflow