Feature Request: average is not debug

[vnp85]

I'm requesting the changing of the class of the average image, from the current bulk debug to a separate optional debug or even mandatory debug output, thus maintaining the output folder structure.

While the edge, recon and tilt, not to mention raw are truly debug images, with little to no use if everything goes well, the average image, with the marking on it, is key to calibrating the pixelOffset-wavelength correlation, when generating a cube for example.

The rationale is that at the moment there are two choices:

• running a "simple" processing, with debug images enabled, to obtain the average then running ImageMath with debug images disabled
• running only an "imageMath" processing, with debug images enabled, to have the average, but accepting gigabytes upon gigabytes of images getting written as a byproduct

Thank you for considering

[vnp85]

Further rationale:

beyond the most obvious fraunhofer lines, it is not obvious where the zero offset will land. In blue, there are several cases that mandate the presence of the marked average image, like this one for the blue helium at 4471.5

[melix]

I don't see why raw should be considered a debug image. One could argue it's less of a debug image than the average one, because it corresponds to a real reconstruction, it's the raw image, unprocessed, no geometry correction. The average image on the other hand is an implementation detail we could get rid of if another technique to figure out the line was used.

[vnp85]

As long as now (that there is no surprise-stretching) img(0) corresponds to raw_0_00, I think this aspect is settled.