Crop into CIF (or SVGA) in-camera instead of scale to improve clarity

[Physics-Dude]

This PR simply crops the CIF base image (400x296) to the middle 240x240 pixels in-camera.

• Changes setCameraResolution to use set_res_raw instead of set_framesize
• Crops the image without scaling/distorting the frame. (increases zoom/decreases FOV by 20%)
• Includes other examples I tested with awesome results.

Cropping into a CIF image should always be done instead of scaling/distorting it (as with what FRAMESIZE_240X240 does). You get weird aliasing or non-uniform binning when you scale from 296 to 240. The "cost" of being 20% zoomed in is more of a benefit I'd wager too since the edge of your lens is never going to be as sharp. This is like upgrading a 160º camera to a crisp 130º.

[Physics-Dude]

Here is a GIF showing some of the above examples on a grid and circle. These were all taken with an old generic ESP-CAM with 160º camera over home wifi.

[Physics-Dude]

Here is a simulated example of the quality difference between scaling CIF to 240 and simply cropping a 240 window out of a CIF. The pixel ratio between the sensor array and image sent to the PC becomes 1:1.

[Physics-Dude]

Playing with some new HW and thought I'd get a IRL side by side image. Left uses a 240px crop within CIF field (default on this PR). One may argue eyelash fidelity is a bit less noisy too.