Reset partial record state after skipping all requested records

[jonded94]

Which issue does this PR close?

• Closes Error "Not all children array length are the same!" when decoding rows spanning across page boundaries in parquet file when using RowSelection #9370 .

Rationale for this change

The bug occurs when using RowSelection with nested types (like List) when:

1. A column has multiple pages in a row group
2. The selected rows span across page boundaries
3. The first page is entirely consumed during skip operations

The issue was in arrow-rs/parquet/src/column/reader.rs:287-382 (skip_records function).

Root cause: When skip_records completed successfully after crossing page boundaries, the has_partial state in the RepetitionLevelDecoder could incorrectly remain true.

This happened when:

• The skip operation exhausted a page where has_record_delimiter was false
• The skip found the remaining records on the next page by counting a delimiter at index 0
• When a subsequent read_records(1) was called, the stale has_partial=true state caused count_records to incorrectly interpret the first repetition level (0) at index 0 as ending a "phantom" partial record, returning (1 record, 0 levels, 0 values) instead of properly reading the actual record data.

For a more descriptive explanation, look here: #9370 (comment)

What changes are included in this PR?

Added code at the end of skip_records to reset the partial record state when all requested records have been successfully skipped.

This ensures that after skip_records completes, we're at a clean record boundary with no lingering partial record state, fixing the array length mismatch in StructArrayReader.

Are these changes tested?

In b52e043 I added a test that I validated to fail whenever I remove my fix.

  Bug Mechanism                                                                                                                                                                                             
                                                                                                                                                                                                            
  The bug requires three ingredients:                                                                                                                                                                       

  1. Page 1 (DataPage v1): Contains a nested column (with rep levels). During skip_records, all levels on this page are consumed. count_records sees no following rep=0 delimiter, so it sets               
  has_partial=true. Since has_record_delimiter is false (the default InMemoryPageReader returns false when more pages exist), flush_partial is not called.
  2. Page 2 (DataPage v2): Has num_rows available in its metadata. When num_rows <= remaining_records, the entire page is skipped via skip_next_page() — this does not touch the rep level decoder at all,
  so has_partial remains stale true from page 1.
  3. Page 3 (DataPage v1): When read_records loads this page, the stale has_partial=true causes the rep=0 at position 0 to be misinterpreted as completing a "phantom" partial record. This produces (1
  record, 0 levels, 0 values) instead of reading the actual record data.

  Test Verification

  - With fix (flush_partial at end of skip_records): read_records(1) correctly returns (1, 2, 2) with values [70, 80]
  - Without fix: read_records(1) returns (1, 0, 0) — a phantom record with no data, which is what causes the "Not all children array length are the same!" error when different sibling columns in a struct
  produce different record counts