GIF animation/layers smarter compression

[GoogleCodeExporter]

GIF files have a single canvas size, and then each included bitmap has its own 
dimensions as well as offsets relative to the canvas.
Files saved by Grafx2 could be made smaller by checking the smallest rectangle 
that isn't made of "background color"; or for animation layers beyond the 
first: the smallest rectangle that has any changes compared to previous frame.
The difference wouldn't be important for single images, but for large 
animations it can be very significant : One of the animations by Stickman was 
5x bigger when re-saved by Grafx2.

Checking this would take more time, but then there would be less data to 
encode, and with the reduced disk access it might even save some time. The time 
gain would be meaningful for the safety backups that happens every minute, when 
saving huge files.

Original issue reported on code.google.com by yrizoud on 9 Oct 2012 at 11:24