After spending some time writing some code to implement my own phase detector, I can confirm that Simulink library's phase locked loop block does indeed have a 270 (or 90 degrees) lag for the VCO signal (relative to the incoming signal) when locked. While it is certainly possible to have phase locked loops that have 90 degrees phase difference when locked, I'm also thinking why Simulink doesn't supply a 'normal' version that simply locks with a ZERO degrees phase difference.
Also, I find it strange how the Matlab/Simulink 'help' documentation doesn't bother to indicate this 90 degrees out of phase locking. If it were me, I would have just included a 'normal' phase locked loop block in the Simulink library, and would also just provide a very simple and straight-forward phase locked loop example.
I'll upload my phase locked loop example for others to follow. It's a Simulink model done in Matlab 2013b.
