Update msgsteiner

[agitter]

These changes relate to running Forest with msgsteiner-1.1 instead of msgsteiner-1.3. The new version fixes the bug described in #17 and can identify lower cost optimal forests. For our A549 example, it returns the same nodes but replaces two edges. This yields a lower overall edge cost.

Specific changes include:

• Update the README to recommend msgsteiner-1.3 and link to the new version (still link to the old site for the licence terms)
• Update Travis CI testing to use msgsteiner-1.3
• Write the objective function score to the info file to verify msgsteiner-1.3 improves the forest
• Update some of the unit tests to test that the objective function is computed correctly
• Update the reference files for the A549 integration test to match the msgsteiner-1.3 output

@AmandaKedaigle Can you please update http://fraenkel-nsf.csbi.mit.edu/omicsintegrator/ to match the changes I made in the README? Specifically, recommending msgsteiner-1.1 and linking to the new code on Alfredo's site.

@aabaker99 I updated some of your unit tests to check the objective function value. This was primarily to verify my calculation of the objective function in forest.py. I don't think we need to verify this in every unit test.

I'll merge this at the end of the week if there are no comments.

[AmandaKedaigle]

I updated the website to reflect the change to 1.3 and new link to msgsteiner