Add new filters

[dkirkby]

Add filter curves for MzLS-z, BASS-g and BASS-r:

Fixes #11.

[dkirkby]

@moustakas @sbailey please check the following:

• Do we want to trim some of the values below 1e-4, or keep everything? (the costs are performance and convenience)
• Any corrections / additions for the descriptions in the new ecsv files (e.g., here).

[dkirkby]

Travis tests are failing because they span numpy versions with different whitespace in their formatted output and I don't know how to tell doctest to do the right thing (even after trying #doctest: +NORMALIZE_WHITESPACE).

For example:

>>> print((result['flux'][:5]))
Expected:
    [ 0.5  0.5  0.5  0.5  0.5]
Got:
    [0.5 0.5 0.5 0.5 0.5]

Unless @weaverba137 knows the answer, I will remove these doctests (which seems a shame).

[weaverba137]

@dkirkby, what specific lines in what files have this issue?

[weaverba137]

PS, I've had success with +NORMALIZE_WHITESPACE. I've also had success using print(). However, I don't think I've done both at the same time.

[dkirkby]

@weaverba137 I tried both print(...) and +NORMALIZE_WHITESPACE (individually).

The 3 lines causing problems are speclite/filters.py #564 and speclite/redshift.py #40 and #56. A sample travis failure log is here.

[sbailey]

Let's go ahead and keep the full wavelength range as you currently have committed. The tails don't seem excessive. Comments in the metadata seem fine (though I haven't checked in detail).

[weaverba137]

I remember a fix now: make numpy print strings using 1.13 format. See https://github.com/weaverba137/pydl/blob/master/pydl/__init__.py#L42.