[ozenyunus35]

Follow-up to #15625, as suggested in #15625 (comment).

extract_strings_from_input had its own copy of the offset formats, which is how the -S -Ax width drifted from the normal dump. It now prints the offset through InputOffset, so both use the same formats.

• InputOffset implements Display with the formats that format_byte_offset used. format_byte_offset now returns self.to_string(), so its callers and unit tests are unchanged.
• -S writes the offset with writeln!(writer, "{offset} {string_content}"). With -An it still prints only the string. I first used format_byte_offset() there, but the String it allocates for each string made od -S4 about 2% slower on a 400k-string file. With Display there is no extra allocation, and the time is the same as on main.

There is no change in output. All test_od tests and the uu_od unit tests pass. od -S with -Ad, -Ax, -Ao, -An, -j and -N, plus a few normal dumps, give the same output as GNU 9.12. cargo bench -p uu_od and timings of od -S4, od -Ax -S4 and od -Ax -tx1 with a release build show no change against the merge base beyond run-to-run noise.

I used an AI assistant to help draft parts of the code; I reviewed the changes and ran the tests myself.

[github-actions]

Binary size comparison:

Individual binary size comparison VS main (threshold: >=5% AND >=4 KB).

Total size of compared binaries: 155.04 MB (-308 KB, -0.19%)

Significant per-binary changes:
  date     2.61 MB ->    2.30 MB  (-316 KB, -11.81%)

[sylvestre]

Thanks for your PR