nintendo publishes around 20-24 games in a given year across the 2-4 systems they'll support at any given moment (OH 2012 WHAT A WONDROUS TIME). when there are fewer platforms, or platforms they're focusing on less, those tend to suffer as far as releases go. see the gba from 2001-2007, the gc from 2005-2006, the ds from 2008-2012, the 3ds's 2014, and so on and so forth. it takes so much time and money to train everyone on two new pieces of hardware every 4 years or so and they have to keep hiring more people and take more time to do it whenever it happens.
quality is subjective. i'd argue it's been pretty consistently good since 2004, the output on the ds notwithstanding. quantity is fairly easily measured, and as you point out the lack of third-party releases makes these platforms look particularly barren in terms of lineup.
ideally a multi-device approach using the same library would ease the learning curve (so there's only one platform to get to know), reduce gaps in release dates, and allow nintendo to come up with unique and profitable hardware that they see fit. for instance, imagine if the wii u this year had its current releases plus what the 3ds had. it would look like this:
february:
bravely default
dkctf
march:
yoshi's new island
april:
disney magical world
may:
mario kart 8
mario golf: world tour
kirby triple deluxe
june:
tomodachi life
...instead of this:
february:
dkc tf
may:
mario kart 8
it would also prevent similar games from coming out somewhat close to one another, like nsmb2 and nsmbu, two very similar smash bros. games (which seems increasingly like a bad idea), or even super mario 3d land and super mario 3d world. there would be a little more breathing room and time for experimentation.
ideally, of course.