Interface for a noise model

[aymgal]

This PR is about implementing an interface for optimizing noise properties during lens modelling (issue #35).

[coveralls]

Pull Request Test Coverage Report for Build 11853678423

Details

• 36 of 51 (70.59%) changed or added relevant lines in 3 files are covered.
• 2 unchanged lines in 1 file lost coverage.
• Overall coverage increased (+0.1%) to 21.503%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
herculens/Analysis/plot.py	0	5	0.0%
herculens/Instrument/noise.py	28	38	73.68%

Files with Coverage Reduction	New Missed Lines	%
herculens/Instrument/noise.py	2	70.87%

Totals
Change from base Build 11707283366:	0.1%
Covered Lines:	1594
Relevant Lines:	7413

---

💛 - Coveralls

[aymgal]

Hi @CKrawczyk , I'd appreciate any comments you may have about this feature, before merging it (since we discussed about it earlier this year) :)

Essentially, you can use it in a numpyro model as:

background_rms = numpyro.sample('noise_sigma_bkg', dist.Uniform(low=0., high=1.))
model_variance = lens_image.C_D_model(model_image, background_rms=background_rms)

and you could also provide a noise "boost map" as a 2D array that locally scales the variance, as C_D_model(..., boost_map=XXX).

I also updated the starting guide to showcase the feature.

[CKrawczyk]

These changes look great, very helpful for how we are using the code.

[aymgal]

Great @CKrawczyk 👌