I've tested vblank events to at least make sure we're firing them off from
the driver, and am currently in the process of fixing one remaining vblank
ref leak. It -should- be fixed now, but I have yet to actually run this
code in my VM and am currently pushing this so there's something to look at
for my team status report :).

Things to split out of this commit:
* Random misc. cleanups
* One safety addition to the C side of DRM (specifically, if one CRTC
  supports vblanks it does not make sense for others to not have things
  like enable_vblank/disable_vblank not implemented - so check for that)
* More hrtimer WIP
* Some additions to Ktime