Return more than one rwalk sample on each iteration

Currently, each parallel process returns a single sample point from rwalk. By returning X fair draws from the rwalk mcmc, this should be equivalent to increasing the number of parallel processes by a factor of around X.

I've started by looking in analysis.py. On the surface, it seems easy enough. If we wanted to return, say, half of the samples we'd just do (line 790):

# If the act is finite, pick randomly from within the chain
    if np.isfinite(act) and int(.5 * nact * act) < len(u_list):
        n_draws = max(1,int(len(u_list)/2.)) # take up to half of the samples
        idx = np.random.randint(int(.5 * nact * act), len(u_list), size=n_draws)
        u = list(np.array(u_list)[idx])
        v = list(np.array(v_list)[idx])
        logl = list(np.array(logl_list)[idx])

However, this fails down the line

Traceback (most recent call last):
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/bin/parallel_bilby_analysis", line 33, in <module>
    sys.exit(load_entry_point('parallel-bilby==0.1.4', 'console_scripts', 'parallel_bilby_analysis')())
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/lib/python3.6/site-packages/pkg_resources/__init__.py", line 488, in load_entry_point
    return get_distribution(dist).load_entry_point(group, name)
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/lib/python3.6/site-packages/pkg_resources/__init__.py", line 2861, in load_entry_point
    return ep.load()
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/lib/python3.6/site-packages/pkg_resources/__init__.py", line 2461, in load
    return self.resolve()
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/lib/python3.6/site-packages/pkg_resources/__init__.py", line 2467, in resolve
    module = __import__(self.module_name, fromlist=['__name__'], level=0)
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/lib/python3.6/site-packages/parallel_bilby-0.1.4-py3.6.egg/parallel_bilby/analysis.py", line 421, in <module>
    for it, res in enumerate(sampler.sample(**sampler_kwargs)):
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/lib/python3.6/site-packages/dynesty-1.0.1-py3.6.egg/dynesty/sampler.py", line 782, in sample
    u, v, logl, nc = self._new_point(loglstar_new, logvol)
  File "/fred/oz170/rsmith/anaconda3/envs/pbilby_exp/lib/python3.6/site-packages/dynesty-1.0.1-py3.6.egg/dynesty/sampler.py", line 393, in _new_point
    if logl >= loglstar:
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

I think the amount of code changes should be minimal. It's probably just a matter of appending lists correctly.