Appion

There are two ways to go about implementing gautomatch:

• Let gAutomatch control the output/processing
• gAutomatch does cross-correlation and Appion processes the CC map and picks particles for each image.

Since the program is GPU-based we should probably let gAutomatch do most of the work as it should be faster (image filtering, multiple templates, etc.), but that gives us less control over the picking process and then we would still have to generate the summary image, which takes some time, because we need to filter the image...

Maybe, we should have two versions one where Appion processes the CC map and picks particles for each image, then we can do better debugging/fine tuning of the process and another one build for raw speed...

This implementation does not have test result for display in appion page when testing.