CBC convergence fixes (Fisher + Heterodyne + Tidal proposal)

Major changes

• Compute reference heterodyne for every temperature instead of using the cold chain as reference for chains > 5
• Change Fisher matrix computation main CBC MCMC. Now more similar to BW's Fisher matrix with:
  • Two-step central differencing instead of one
  • Reduce diagonal stabilizer to 1e1 instead of 1e5
  • Set epsilon size for each parameter and use Likelihoods to inform size, instead of one epsilon fits all parameters
  • Added range checks in Fisher for spins, masses and distance, because hot chains for BBHS could propose too large epsilons near the edge leading to negative masses/distance or abs(spins) > 1 for the Fisher steps
• Added a new proposal in the CBC MCMC for runs with tidal parameters that jumps uniformly in the tidal parameters, but draws the other parameters from history. This because even with the adjusted Fisher, the jumps in the tidal parameters remained very small. Including this proposal led to the tidal params being able to break out. Perhaps still worth looking into why the other proposals for tidal parameters keep suggesting such small jumps.

Minor changes

• Intrinsic burnin cleaned up and proposal ratio adjusted to speed up signal finding
• Intrinsic burnin slightly shortened
• Make jiggle proposal base steps larger for tidal parameters
• Added names to CBC files
• Clean up comments/prints

TODO's

• Test BNS runs
  • GW170817
  • Injection
• Test BBH runs: Stop runs from stopping when the Fisher computation wants a parameter that the waveform can't handle (i.e., masses/distance < 0 and spins out of -1,1 range)
• Clean up comments/prints