Filter func tests

[daviesje]

This PR combines the three filtering functions present in v4-prep, and makes some tests for the filter functions which:

• ensures that each filter doesn't change the mean of the box (or changes it the right amount)
• ensures that the profile produced by a single cell matches what we would expect from a filtered delta function, within some tolerance

Here are the test plots produced for each filter with R=10:
tests.test_filtering.py--test_filters[2-10].pdf
tests.test_filtering.py--test_filters[1-10].pdf
tests.test_filtering.py--test_filters[0-10].pdf
tests.test_filtering.py--test_filters[4-10].pdf
tests.test_filtering.py--test_filters[3-10].pdf

Still TODO:

• fix the first test for the exponential filter
• set a reasonable passing criteria for the second test

[steven-murray]

Thanks @daviesje -- I have a couple of suggestions that might help, but let's see

[steven-murray]

can we write a low-level wrapper function for this in the main code?

[steven-murray]

In fact, if instead we write the filter_box function as a low-level wrapper, we could perhaps write the test_filter function directly in python?

[daviesje]

Would the second option not require us to expose a bunch of fftw functions to the python wrapper?

[steven-murray]

Not necessarily -- we could just use np.fft or even pyfftw to do the FFT stuff in the wrapper function

[daviesje]

Wouldn't we be testing different funcitons at that point?

[daviesje]

If you're okay with it, I would prefer to write wrappers for all the testing functions together in a separate PR

[steven-murray]

That's fine with me

[daviesje]

So the tests now pass (at least the filter tests). I've significantly widened the tolerances on the pixel-to-pixel test, and included a radially binned test in order to mitigate some problems with sharp real-space filters. Below are all the plots from these tests

tests.test_filtering.py--test_filters[4-1.5].pdf
tests.test_filtering.py--test_filters[4-5].pdf
tests.test_filtering.py--test_filters[4-10].pdf
tests.test_filtering.py--test_filters[4-20].pdf
tests.test_filtering.py--test_filters[0-1.5].pdf
tests.test_filtering.py--test_filters[0-5].pdf
tests.test_filtering.py--test_filters[0-10].pdf
tests.test_filtering.py--test_filters[0-20].pdf
tests.test_filtering.py--test_filters[1-1.5].pdf
tests.test_filtering.py--test_filters[1-5].pdf
tests.test_filtering.py--test_filters[1-10].pdf
tests.test_filtering.py--test_filters[1-20].pdf
tests.test_filtering.py--test_filters[2-1.5].pdf
tests.test_filtering.py--test_filters[2-5].pdf
tests.test_filtering.py--test_filters[2-10].pdf
tests.test_filtering.py--test_filters[2-20].pdf
tests.test_filtering.py--test_filters[3-1.5].pdf
tests.test_filtering.py--test_filters[3-5].pdf
tests.test_filtering.py--test_filters[3-10].pdf
tests.test_filtering.py--test_filters[3-20].pdf

[steven-murray]

Thanks @daviesje this all looks good. Quite wide tolerances, but I think in conjunction with the plots, this is good.