fix: handle invalid byte ranges in calculate_range for single-line files

[vigimite]

Which issue does this PR close?

• Closes calculate_range creates invalid byte ranges for single-line JSON files #19605.

Rationale for this change

The calculate_range function creates invalid byte ranges (where start > end) when reading single-line CSV/JSON files that are split into multiple partitions. This causes an error like:

ObjectStore(Generic { store: "S3", source: Inconsistent { start: 1149247, end: 1149246 } })

When find_first_newline doesn't find a newline (single-line file), it returns the remaining file length. This causes start + start_delta to exceed end + end_delta, creating an invalid range. The current check only handles range.start == range.end, not range.start > range.end.

What changes are included in this PR?

1. Added an early termination check after computing start_delta: if the first newline after start is beyond the partition boundary (start + start_delta > end), return TerminateEarly since no complete records exist in this partition.

2. Changed the final range validation from == to >= as a safety net for edge cases.

3. Added a regression test that reproduces the bug with a single-line file split into partitions.

Are these changes tested?

Yes, added test_calculate_range_single_line_file which:

• Creates a single-line JSON file without new-lines
• Simulates partition 2 (middle to end of file)
• Verifies that calculate_range returns TerminateEarly instead of an invalid range

Are there any user-facing changes?

No

[martin-g]

LGTM

[Jefffrey]

Nice pickup, thanks @vigimite & @martin-g