Feature: Provide Cross section library for running tests

[yardasol]

Background and motivation
See #72 for background.
The TL;DR is that we currently do not specify in the docs any particular cross section library that is necessary for the tests, even though the tests themselves are set up to pass for a particular library. We should do two things:

1. Specify in the docs that users need to download and set up a specific library in order for the tests to pass
2. Provide the library to the users if it requires more work than downloading, extracting, and modifying .bashrc.

Description of idea
In #72 I detailed my process for reconstructing the JEFF 3.1.2 data into a library that can be used with Serpent. I suggest we use this library and put it somewhere on the saltproc website, or link to it straight from the repo.

Implementation details
We need to add a section in the docs somewhere that talks about how to set up the library, and specify that it is required for the test suite to pass.

We also need to host a downloadable archive of the library somewhere. @munkm any ideas?

Potential snags
I'm still working through getting the integration tests up and running again, so the library isn't actually ready yet. Once #72 is closed, we can work on making a PR to resolve this issue.

[munkm]

So, for the openMC tests we can assume that users can download the cross section libraries from the openMC docs pages. If somebody does have a serpent license, we can also assume they also have a cross section library that came distributed with Serpent. Rather than distributing cross section data (which I think would get us in some murky waters and would require us to verify any person downloading the data from our pages to have a license) I think what we should do is recommend that users run the tests with a specific library, but make sure our tests have some "fuzziness" on the acceptable answers to account for using different libraries. Things like numpy's assert_almost_equal rather than an assert_equal would be relevant here.

[yardasol]

So, for the openMC tests we can assume that users can download the cross section libraries from the openMC docs pages

Agreed

If somebody does have a serpent license, we can also assume they also have a cross section library that came distributed with Serpent.

Ditto

Rather than distributing cross section data (which I think would get us in some murky waters and would require us to verify any person downloading the data from our pages to have a license)

I agree that we shouldn't distribute the cross section data included with Serpent as that would in violate of the license with RSICC. What I was trying to communicate was that anyone can do what I did to the JEFF 3.1.2 library (as it is publicly available and is not included with serpent at all) to make it usable with serpent, so why make users do all that work if we can just provide the files to them instead. However I hadn't considered whether or not we are even allowed to do that since OECD provides the JEFF libraries, but the OpenMC people host their own data libraries so I figured we could as well.

I think what we should do is recommend that users run the tests with a specific library, but make sure our tests have some "fuzziness" on the acceptable answers to account for using different libraries.

I think this is the easiest solution. It will mean we need to modify the integration test input files, but that is probably less of a hassle than trying to set up a hosting page for a cross section library.

[yardasol]

Closing this issue as we aren't going to be hosting our own cross section library after all. See #78 and #79.