Split into operational classes

[elliotwoods]

We need 3 seperate operational classes:

1. Grabber acts like ofVideoGrabber (has all usual getters, inherits lots of things, can be drawn to screen, etc)
2. EventGrabber gives purely frame callbacks (no texture, no getpixels, etc). Gives event calls when frame arrives from capture thread.
3. BlockingGrabber is for 'wintin thread' operation (whenever you ask for frame, it blocks until new frame is ready)

Then seperately we have the Device interface which is implemented for each camera API.

Then we have DeviceControl interface which includes things like:

• ROI
• Triggering
• PIxelMode

Device inherits interface DeviceControl.

For user at runtime:

1. Grabber implements DeviceControl
2. EventGrabber gives a DeviceControl on every callback
3. BlockingGrabber implements DeviceControl

The user then uses syntax like:

Grabber grabber(new ofxXimea(deviceID));

or perhaps

Grabber grabber;
grabber.init<ofxXimea>(deviceID); ///< this calls an inline templated function