ItIsActuallyFindingFacesICantBelieveIt

This is the first commit where I'm satisfied enough to say the code is actually finding faces automatically. I still have it set up where the user needs to click for the initial location of the face (rather than an automatic global face finder like Viola-Jones), but the initial estimate can be fairly far off and the ASM will still converge on the face.

Time to start fine-tuning.