Add new halo models to convert_concentration

[hsinfan1996]

Add support for the Einasto and Hernquist profiles, although the Einasto profile itself uses the equation for NFW when getting the alpha parameter from cosmology. Now the function is in pure python, and it runs slower. For comparison, converting one concentration value takes 5 μs before vs 270 μs now and converting an array of 10 concentration values takes 6 μs before vs 600 μs now. I will close this PR if this is unacceptably slow.

[hsinfan1996]

~I cannot seem to fix the tests, but looks like~ it is swig related. In swig-generated ccllib.py, if numpy.shape(c_in) != (nout,): was placed under cosmology_compute_sigma, where it should not be.

FIXED: %feature("pythonprepend") was not added at the end of ccl_musigma.i, which affected the next file after ccl_mass_conversion.i was removed. Not sure if it had been intended.

[coveralls]

Pull Request Test Coverage Report for Build 5571318719

• 18 of 18 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.005%) to 97.434%

---

Totals
Change from base Build 5553290862:	0.005%
Covered Lines:	5657
Relevant Lines:	5806

---

💛 - Coveralls

[damonge]

Humm, although this is not something that is currently heavily used, I would rather not replace something in the code by a version that is so much slower. I can see two options:

• Leave the existing NFW code in place for NFW, and use the method you implemented for the other two profiles.
• Even better, implement the other two in C. A quick look at the code tells me it's just a matter of replacing the function whose root we need to find, so perhaps it's not so complicated.

[hsinfan1996]

Not planning on touching c-level files. Since CLMM already had this functionality, will close this PR.