feat: add boundary utility functions for aligned byte fetching in JSON

[Weijun-H]

Which issue does this PR close?

• Closes #.

Rationale for this change

When scanning partitioned JSON files on remote storage systems (HDFS, S3), the current calculate_range() implementation causes severe read amplification:

Current behavior:

• For each partition boundary, calls find_first_newline(start, file_size)
• Requests byte range [start..file_size) from object store
• Observed 4-7x read amplification in production (reading 278MB-1084MB to find newlines)

Why this is problematic:

1. Remote storage systems have high latency for range requests
2. Reading unnecessary data wastes network bandwidth
3. For large files (100MB+), boundary checks dominate scan time
4. The newline character is typically within a few KB from the boundary

What changes are included in this PR?

Implements get_aligned_bytes() with efficient in-memory boundary alignment:

Start boundary alignment:

• Fetch only [start-1..end] range
• If start == 0, use as-is
• Else check if bytes[0] (position start-1) is newline
• If yes, start from start; if no, scan forward in memory for first newline
• Return None if no newline found in fetched range

End boundary alignment:

• Fast path: If end >= file_size or last byte is newline, return immediately (zero-copy)
• Slow path: Extend in small chunks (4KB default) until newline found
• Pre-allocate capacity to reduce reallocations

Key optimization:

• 95%+ of cases hit the fast path (boundaries already aligned)
• Uses Bytes::slice() for zero-copy when possible
• Only allocates Vec when extension is actually needed

Are these changes tested?

Yes

Are there any user-facing changes?

No

[martin-g]

Why don't you use u64 for these ?
Here you cast them to usize and at https://github.com/apache/datafusion/pull/19687/changes#diff-d0a9c47dbd0bdb20995b4a685f0f7551bebf22287035c99636f2b98013f203b0R52 you cast them back to u64.
The castings here may lead to problems on 32bit systems.

[Weijun-H]

update in c188abc

[martin-g]

Please also address the clippy errors - https://github.com/apache/datafusion/actions/runs/20811221444/job/59776119936?pr=19687

[Weijun-H]

fixed in 5eec399

[martin-g]

It would be good to add a check that scan_window is bigger than 0, otherwise here the get_range() below will use an empty range.