I can see how niche game-modes could easily spring up from the implementation of community created maps, and that it could take players away from the "main game" - same applies to custom server hosting/configurations. However my mind more goes to the thought of, well if these mode(s) are becoming more popular than the game itself, are they..
a) More fun than the actual game
or b) Made too accessible to the average player
Either way it's a problem, but I would say it's more in the hands of the developer to balance it out and setup the user flow in such a way that the focus stays in the right spot. Cutting off this facet of creativity would be a sure-fire way of preventing this for sure, but I do think there are potential options for making it work.
Going back to the TF2 example, they have made it so that the main part of the matchmaker will only consider servers running official maps only. It has the option to straight up join the best server (based on ping), or just list the servers and their pings/player count so you can choose the map you want. It then has an added area of 'beta maps' where servers running the official 'in progress' maps are listed -
<hint> this could also be a cool place to show off any community maps that have potential to be officially integrated into the main rotation
Then completely separate to this process is the flat server list which has every single server, both official and non-official, with host information, map information etc, which would then allow players to choose any server, which is where the community maps and/or modes would stem from.
That's just where my mind goes on the subject. I know many competitive shooters do opt for the approach you describe and it has worked out great, however I can also see the great potential that arises from users being enabled to put countless hours into possibly creating and testing the "next awesome map", that devs alone would never have the time for.