updated lalinference to sample in 4-piece polytrope EOS parameters

Hi,

This patch is for implementing a new way to do EOS PE into the MCMC sampler. Here, we model the EOS as a 4-piece polytrope (using functionality in LALSimulation), and we then sample in the 4 parameters of this EOS model. Flat priors are assigned to each parameter, but acausal regions of the parameters space are also zeroed out in the prior. As an option, you can fold max-NS-mass observations into the prior too.

Note: There are a couple of small changes to LALSimulation, which is where the code for constructing the EOS models lives. These are mostly fixes to get around EOSs that were failing. This used to be a patch on Redmine, and Jolien responded, "The LALSimNeutronStar*.c changes seem okay."

I am currently running the following on CIT:

mpirun -n 8 lalinference_mcmc \
        --resume \
        --ntemps 8 \
        --ifo H1 --ifo L1 --ifo V1 \
        --H1-cache LALSimAdLIGO --L1-cache LALSimAdLIGO --V1-cache LALSimAdVirgo \
        --H1-flow 30 --L1-flow 30 --V1-flow 30 \
        --inj Pop1.xml.gz --event 0 \
        --trigtime 10 \
        --psdlength 3200 --psdstart 20 \
        --srate 4096 \
        --seglen 100 \
        --dataseed 10 \
        --randomseed 437877036 \
        --approximant TaylorF2threePointFivePN \
        --amporder 0 \
        --neff 4000 \
        --disable-spin \
        --4PolyEOS \
        --inj-logp1 34.495 --inj-gamma1 3.446 --inj-gamma2 3.572 --inj-gamma3 2.887 \
        --ns-max-mass 1.97 \
        --distance-max 300 \
        --0noise

The path to the working directory is here:

/home/lwade/4piece_test/master_patch

This has only been running for a little over a day, but I can update when it has more samples.

I've been in touch with Vivien Raymond and Ben Farr about this merge request. Let me know if there are any questions/comments.

Thanks, Les