In short there still aren't any standardized rom lists for the mess/console side of things, many popular consoles aren't fully emulated or run really slow even on powerful machines.... so yeah, not much point in using it for that and keep in mind that I am very pro mame/mess.
I love the idea of MESS, I really do, and I totally understand that from a development standpoint it makes sense to go ahead and merge them since they have really been the same thing for quite some time BUT I sure wish the MESS end of things would have been standardized and given the level of polish the MAME end has before they were officially integrated.
Like those special dsp chip games... run perfectly fine on more popular console emulators, even ones that strive for accuracy like bsnes.
It's coming along.
Things like the SNES expansion chips are going to be demanding due to the MAME architecture, they're powerful sub-cpus that need to run in tight sync with the main CPU, MAME has never really been that fast in such cases.
The Software Lists will evolve over time, some are currently very good (things like the Genesis one, which is probably *the* best list of correct Genesis dumps with proper documentation of chip labels etc. when available you'll find) others are works in progress. (things like the Saturn list where we need to work out which images are good / bad by working out if CHDMAN mangled any of them due to .cue/.bin being such a loose standard etc.) or the CPC list which is very useful, but not in any way tidy right now.
For rare Japanese computers / consoles the support is actually the best; for things like the Casio pv1000 and pv2000 it's the best out there. For the recently added handheld electronic games (as long as you have the artwork) we're still the only ones even attempting to do that properly.
We couldn't really "wait until it's all polished to integrate it" because that would be like saying we shouldn't have released MAME until every arcade game was emulated; it's a distant goal, the only way we can get there is to go through the various iterations, evolve, improve, rework etc. and the only way that process works is if we get people onboard helping.
In the end it's free content, you can choose to not use it, or you can choose to use it and have a bit extra. If you don't like it, don't use it, it's like the old 70s b&w games in MAME, most of them hold very very little interest to me, but they exist and hold interest to other people. In the end we emulate hardware, emulating whatever hardware we can emulate makes perfect sense from the point of view of the development of the emulator, it provides extra test cases out the box whereas splitting stuff away into another project because some people might not want it doesn't make sense.
It's a change, it's been coming for a long time (4+ years) so there has been ample warning, but like any change it's going to take people a little while to adapt to it. I expect there will be some moaning, people like something to moan about, doesn't matter if it's mahjongs, casino games, porn games, working on obscure arcade games instead of making their favourite 3d ones fast, or now this. For every person moaning about what we do there are multiple other people appreciating it.
FWIW I think you're being a bit harsh anyway, while we might not be the *best* emulator for a number of the systems we're certainly good enough for many, and things like the vector HLSL make Vectrex a joy, much more pleasing than the competition, even things like our handling of widescreen displays properly makes MAME a better choice than some of the single machine emulators I've used where they still seem to insist on stretching a 4:3 display over the entirety of the widescreen monitor with no way to correct the aspect. The fact is I could pick a game like Raiden, and run every version of it to a playable degree with the new MAME says a lot (
http://mamedev.emulab.it/haze/2015/06/01/what-to-do-in-one-six-two-part-1/ ) and shows just how much more capable this project is in this form.