Yes the project has somewhat merged. However, you can play only arcade games if you wish by simply just downloading and using MAME. Mess ISN'T required to play. A complete MAME 0.186 merged romset will set you back 57.623GB of space. A non-merged complete 0.186 set will cost you 124.014GB of space. CHD files not included. Now you will have bios files, and other miscellaneous files included in the set that arent "playable" but are part of the set for documentation. It is up to you to determine what you wish to keep/discard should you want to do that.
You dont need every console/game on your HD. But yes you do need certain roms/romsets to verify a game(in other words to make sure you have every needed file for it to run correctly). As we cant discuss where to acquire such things for legal reasons, you'll have to seek that out across the vast the universe that is the innertubes. Good Luck, have fun and GAME ON!