Don't confuse the regex you use with the filename with the logic Calibre has internally to decide whether to treat two books as duplicates for automerge purposes. Automerge has some "fuzziness" in its comparison of book titles, stripping a whole bunch of characters like brackets, punctuation, title sort characters like "the, a, an" etc. It is certainly not an "exact match" experience.
It will not throw away numeric values (but will the periods separating them). You might "get lucky" and provided every book you import has enough different characters you get it to do what you want. However you are absolutely playing with fire with this, and as the saying goes you may get burned.
It has its purposes - the best usage of it imho is when you have another
format of an existing book in your library that you want to add. Say you have a book record in MOBI format, and now you get an EPUB version from somewhere. For that purpose AutoMerge is brilliant.
However trying to use Automerge in combination with bringing in multiple versions of epubs sounds "wrong" to me. You might get lucky, you might end up with a mish-mash mess.
