This is such a BS answer and to be honest a real disappointment, especially coming from a moderator here I respect 
Plex (no one else) has decided to pull metadata from multiple places, its their code after all that pulls these dups. Plex consolidates that data wrongly, incorrectly pulling multiple entries into one. Yes the data flows from tvdb, but plex has decided within their code logic to pull multiple entries including the empty one left as a husk as you correctly pointed out.
This whole consolidation approach is badly designed with what seems like only basic level of validation from the sources they scrape from. It should not matter how many dups exists, plex could (and should) be improved to validate things better, pick the best and should NEVER have dup ids. Sure, plex don’t even clean up ids when they are removed as is pointed out in this thread, which btw is still an issue so your suggestion of deleting the tvdb entry has no guarantee of fixing this!
Regardless, do with it what you want, I’m done trying to help the community and fix plexs many metadata problems. Slán!
