head: distinguish read and write errors in copy loop

[gabrielhnf]

Fixes #12234

Problem

When running head on a file that fails during reading (e.g. /proc/self/mem), uutils was reporting:

head: error writing 'standard output': Input/output error

Instead of the correct:

head: error reading '/proc/self/mem': Input/output error

Cause

io::copy merges read and write errors into a single return value with no way to distinguish them. Since all errors were mapped through wrap_in_stdout_error, a read failure was incorrectly attributed to stdout.

Fix

Replaced io::copy with a manual read/write loop and introduced a PrintError enum to tag errors at the source as either ReadError or WriteError. This distinction is preserved all the way up to uu_head, where read errors get the filename attached and write errors are reported against stdout.

Note

• print_n_bytes has the same issue via send_n_bytes. Fixing it would require introducing a similar distinction inside that function, which felt out of scope for this PR and thus was left for a separate issue.

• Some code paths in print_but_last_n_bytes, print_but_last_n_lines, and the seekable file path in head_backwards_on_seekable_file still map all errors to PrintError::WriteError as a pragmatic compromise. These paths do not exhibit the bug reported in this issue. Left for a separate issue.

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/tail/retry (fails in this run but passes in the 'main' branch)
Skip an intermittent issue tests/tail/tail-n0f (fails in this run but passes in the 'main' branch)
Note: The gnu test tests/unexpand/bounded-memory is now being skipped but was previously passing.
Congrats! The gnu test tests/printf/printf-surprise is now passing!

[oech3]

Err in 1st splice at send_n_bytes is discarded by fallback. So ti should not appear.

Oh Ok. I should split last loop.

[oech3]

CI's log is not useful form me to investivage what utils is missing import. How did you determine the probrematic commit?

[oech3]

main is passing OpenBSD CI. So it looks network error.

[gabrielhnf]

Noticed rustix missing in the logs and connected it to the commit since it was the most recent merge introducing that dependency. If main is passing then you're right, likely a transient network error. Sorry for getting ahead of myself.