install: fix -D data-loss when destination parent is the root directory

[koopatroopa787]

Summary

install -D src /file silently deletes the source file and then
fails, leaving the user with neither the source nor the destination.

Root cause

When the target is a single-component absolute path like /file,
target.parent() returns "/". Before calling create_dir_all_safe,
the code strips trailing slashes from that parent:

while trimmed_bytes.ends_with(b"/") {
    trimmed_bytes = &trimmed_bytes[..trimmed_bytes.len() - 1];
}

For "/" this strips ALL bytes and produces "".
find_existing_ancestor("") maps an empty path to "." (current
directory), so create_dir_all_safe returns a DirFd pointing at CWD
instead of /.

The subsequent unlink_at(cwd_fd, "file") then removes ./file — the
source — rather than any existing destination. The copy step fails
because the source is gone, and the user ends up with nothing.

Fix

Skip the trailing-slash strip when the result would be empty (i.e. the
path was entirely slashes). The root directory "/" is left as-is, so
create_dir_all_safe opens / as the parent fd and later
unlink_at(root_fd, "file") targets the intended destination.

Changes

• src/uu/install/src/install.rs: guard the slash-strip so an
  all-slash path ("/") is kept intact.
• tests/by-util/test_install.rs: regression test verifying the source
  file survives when install -D src /dest fails (non-root execution).

Fixes #13232

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[codspeed-hq]

Merging this PR will improve performance by 4.84%

⚡ 1 improved benchmark
✅ 330 untouched benchmarks
⏩ 46 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	sort_ascii_utf8_locale	16.1 ms	15.4 ms	+4.84%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing koopatroopa787:fix-install-D-root-parent (a999c25) with main (eae191c)}

Footnotes

1. 46 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/date/date-locale-hour (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/tail/retry (passes in this run but fails in the 'main' branch)