generate-domains-blocklist.py overlapping local patterns

[stevehartwell]

The generate-domains-blocklist.py script has code to eliminate domains which are already matched by a (trusted, file-local) glob expression, saving thousands of duplicate entries which might have a measurable performance improvement.

That optimization seems to have stopped working with a change introduced on 2020-12-20.

The change that intended to ignore time-based glob entries identifies them by matching the regex rx_timed = re.compile(r".+\s*(@\S+)?$"). But since the ? allows @\S+ to occur 0 or once, it actually matches all glob entries, so all of the glob expressions are ignored and no duplicates are eliminated.

When I change this expression to: r".+\s*@\S+$", only time-based glob entries are ignored, and the output now includes lines like:
# Ignored due to overlapping local patterns: NNNN
where they weren't being output before.