Build a realistic galaxy sample, with some typical SNR, shape, size

[EiffL]

This issue is to track the definition of a fairly realistic galaxy sample we may wish to use for testing. Choices made here are being implemented in #34

[EiffL]

So, here is my proposal here: a simple parametric dataset based on CFIS-like conditions:

• pixel size 0.187
• Moffat PSF with FWHM 0.65 arcsec, tiny intrinsic shear
• CFIS depth, i.e. 24 mag == 10 SNR
• COSMOS 25.2 parametric galaxies

I've implemented a first prototype of this in #40, with an option to include a shear of 0.02 or not.

Looks like this:

@aguinot does that look good to you? As a toy reference galaxy sample? We of course have only 80000 cosmos galaxies but we can augment them as much as we want with random rotations/reflections + different noise realisations.

[aguinot]

I think this is reasonable. We may want to cut some galaxies at mag > 24.5 since there are going to be dominated by noise. Regarding the sample size I am not too worry at the moment, I don't see the need for a larger sample given our current experiment. We can think more about that later.
Do you have some images of those galaxies to see what they look like?