* fix(filters): separate each match on its own line for source: watches
#4477
A `source:` watch passes append_pretty_line_formatting=False to the filter
helpers, and then keeps the filtered text verbatim instead of running it
through html_to_text(). TEXT_FILTER_LIST_LINE_SUFFIX ('<br>') only works
because Inscriptis turns it into a newline, so for that path no separator was
emitted at all and every match ran into the next on a single line.
Add filter_match_separator(), which picks '\n' instead of '<br>' when the
result will not be passed through Inscriptis, and use it from
xpath_filter(), xpath1_filter() and include_filters() - all three carried the
same condition. The first match still never gets a separator, so no leading
newline is introduced, and the Inscriptis path is byte-for-byte unchanged.
* fix(filters): don't double-space xpath element matches on source: watches
xpath_filter/xpath1_filter serialise whole elements with
etree.tostring(pretty_print=True), which already terminates every match
with a newline. Adding the new source: separator unconditionally put a
second newline between matched elements, so each one landed on its own
line separated by a blank line.
Only emit the separator when the accumulated output does not already end
on a line break, which leaves attribute and text() matches (plain
strings, no newline of their own) separated as before.
Found while producing the before/after output requested on #4492.
---------
Co-authored-by: Yi-111-a <yi-111-a@users.noreply.github.com>
Co-authored-by: Yi-111-a <41898262+github-actions[bot]@users.noreply.github.com>
