number counts test updates

[rmandelb]

Based on some work I just did with the test of the cumulative number counts, e.g., with outputs like this:

https://portal.nersc.gov/project/lsst/descqa/v2/?run=2018-09-10_10&test=ApparentMagFuncTest_HSCz

I have a few suggestions to make:

1. The vertical axis should say what the units are (number/deg^2)
2. The plot title should say the area for the test. If the area is small enough we should be careful not to over-interpret the results (cosmic variance).
3. I would like to propose adding a few other validation points. These are based on pencil-beam surveys, so I'd suggest just points rather than entire curves: i-band N(i<26.5)=150/arcmin^2 from Subaru Suprime-Cam based on Capak+07, N(i<28)=330/arcmin^2 from HDF-N and HDF-S based on numbers given in Capak+07, N(z<24)=20/arcmin^2 from the UDF (https://arxiv.org/pdf/astro-ph/0607632.pdf) and N(z<27)=200/arcmin^2 from the UDF as well.
4. Could we have text file output with the blue curves, to avoid having to read them off by eye?

[rmandelb]

Note, issue posted too early, so I edited to complete it (please read online).

[yymao]

Thanks for suggestions, @rmandelb. They are all good suggestions. 1, 2, 4 are all very easy to implement. 3 should be straightforward too but does require compiling those data points.

@duncandc are you up to help with this? I'm asking you first because you implemented the original test, but no worries if not. I can also do it myself and/or find someone else to do it.

[rmandelb]

For (3) I was referring specifically to the numbers I gave in the first post, so no more compiling should be needed. (I could write an explanatory comment in the code about where they come from.)

[evevkovacs]

@rmandelb @yymao Thanks for the suggestions. I think we should make it standard to have a text file output of the points in the plots for as many tests as possible. I have had a couple of other people comment on how useful they are. The output should also include any cuts that were made and any other relevant information such as the area above. Output columns should be labeled.

[aphearin]

Just chiming in to confirm that it is extremely useful when the tests come with tabulated data points. When available, I use these all the time in local calibrations.

[duncandc]

I think I can take care of this rather easily. For reference, the points for HSC are available as tabulated data.