#35 adding general and simplified jump proposal draws

[siyuan-chen]

Hi all,
regarding issue #35
I have added 3 functions to JumpProposal in model_utils.py(v1.0, there is a bug) / sampler.py(v2.0) to allow for a more flexible user-specified jump proposal draw. In theory those three functions could be even more generalized and can be used to replace most of the existing jump proposal draws in model_utils. The error handling could also be done better. But, it seems to work for me and I will write some instructions below.
draw_from_par_prior: allows the user to specify a list of parameters that should be jumped by drawing from the prior.
draw_from_par_log_uniform: allows the user to specify a dictionary of parameters that should be jumped by drawing from the given log uniform distribution.
draw_from_signal: allows the user to specify a list of signals that should be jumped by drawing from the signal prior.

Let me know if there is any issues. Hopefully, this is useful for people.

Tutorial code:
custom_par_list1 = ['gw_log10_A']
custom_par_list2 = ['jup_orb_elements_0','gw_log10_A']
custom_par_log_dict1 = {'gw_log10_A' : (-18,-11)}
custom_par_log_dict2 = {'J1713+0747_red_noise_log10_A' : (-20,-10)
,'gw_log10_A' : (-18,-11)}
custom_signal_list = ['red noise','dm_gp']
jp = model_utils.JumpProposal(pta)
sampler.addProposalToCycle(jp.draw_from_par_prior(custom_par_list1),15)
sampler.addProposalToCycle(jp.draw_from_par_prior(custom_par_list2),15)
sampler.addProposalToCycle(jp.draw_from_par_log_uniform(custom_par_log_dict1),15)
sampler.addProposalToCycle(jp.draw_from_par_log_uniform(custom_par_log_dict2),15)
sampler.addProposalToCycle(jp.draw_from_signal(custom_signal_list),15)

By the number of lines:

1. add a prior draw for gw_log10_A every 15 iterations
2. randomly choose between gw_log10_A and jup_orb_elements_0 and then add a prior draw every 15 iterations
3. add a draw from a uniform distribution for gw_log10_A between -18 and -11 every 15 iterations
4. randomly choose between gw_log10_A and J1713+0747_log10_A and then add a uniform distribution draw between -18, -11 for gw and -20,-10 for J1713 every 15 iterations
5. randomly choose between red_noise and dm_gp and then add a draw from the prior for the corresponding signal every 15 iterations.