Decode every escape in /proc/self/mounts, not only spaces

[kevin9327]

parse_mounts and device_in read /proc/self/mounts but only turn \040 back into a space. The kernel escapes four characters in these fields (space \040, tab \011, newline \012, backslash \134), and parse_mountinfo already decodes all of them in one pass with unescape_octal. So the same mount can come out differently depending on which table was read:

/dev/sda1 /media/My\040Disk\134040 ext4 rw 0 0

• parse_mountinfo → /media/My Disk\040 (correct)
• parse_mounts → /media/My Disk\134040 before this change, so the point never matches a real path

The same happens for a mount point with a tab or backslash in its name, and for a source with a space (//nas/My\040Share). parse_mounts feeds the removal guard's mount points (read_mount_points), foreign_mounts, volume_root and the volume list, so such a mount point isn't recognized as one there.

This uses unescape_octal for the source and mount point in parse_mounts and device_in, the two fields parse_mountinfo decodes.

Test: mounts_escapes_are_decoded_once, mirroring mountinfo_escapes_are_decoded_once for parse_mounts and device_in. It fails on main (left: "/media/My Disk\\134040") and passes with the change.

cargo test -p disktree-core --lib     # 137 passed; the same 4 failures as on main here
                                      # (trash / one-volume tests that depend on this sandbox's mounts)
cargo clippy -p disktree-core --all-targets
cargo fmt --check

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.