Implement image gradient calculations + GUI elements

Image gradients are implemented via the existing convolution framework. I have added 5 gradient-computing convolution kernels: Central differences, Sobel 3x3, Sobel 5x5, Prewitt and Roberts (details available in headers). These add finite differences gradients with varying amounts of smoothing or averaging.

I have also implemented a toggle on the DetectorWidget toolbar, allowing the user to switch between the image and its gradient.I have added a combo to switch between the various kernels in SubframeIntegrate, but the integration filtering functionality hasn't been added yet. The default kernel is Sobel 3x3, this is what is used to compute the gradient on all widgets other than SubframeIntegrate, where the kernel can be changed.

To do: impement gradient perpendicular to image plane, obviously can't be done with convolution. Include gradient in integration as a means of filtering peaks with inhomogeneous background.