Preprocessed file count underflow

[cebtenzzre]

Often the preprocessing step is so fast that rmlint never actually prints its final statistics, so you get this:

▕░░░░░░░░░░▏                     Preprocessing (reduces files to 0 / found 0 other lint)

even when files/lint are found. This is fixed by https://github.com/cebtenzzre/rmlint/commit/1903b03df777d6121bd48b4bac7b2b9e6a0cbdaa.

With that fix, this issue is fairly easy to reproduce with a small file and a large minimum size:

$ mkdir a
$ echo foo >a/foo.txt
$ rmlint -o progressbar -o summary -T df -s 1G a
▕░░░░░░░░░░▏                 Traversing (0 usable files / 0 + 0 ignored files / folders)
▕░░░░░░░░░░▏  Preprocessing (reduces files to 18446744073709551615 / found 1 other lint)
▕░░░░░░░░░░▏         Matching (0 dupes of 0 originals; 0 B to scan in 0 files, ETA: ...)

==> 0 file(s) after investigation, nothing to search through.

Without '-s', rmlint will report reduces files to 0 / found 1 other lint which is still wrong but less obvious. This is different from #461 because this issue applies to -T df, and underflows the count in the preprocessing progress and not the duplicate count in the summary.