The best advice I can give you is to do the pdf->epub conversion as cleanly as possible, preserving the text. Then take the epub and open it in Sigil to do the regex work - You can use the latest Sigil beta which has a nice new regex engine.
There should not be a space with that replacement, however if you are replacing it with a space, or the following line starts with a space character, perhaps using something like -<br(\s*/?)>\s* will better match. In either case I would suggest doing work like this outside of Calibre itself. While it may seem like a bit of extra work, it often saves a lot of time in the long run and will get you the results you're looking for.
|