This updates get_input_data_array. First, the list of possible search
parameters is defined as a list of strings in the class itself avoiding
repeated definitions of get_input_data_array. Second, is `args.clean == True`
then the input_data is not generated as an array, but a generator. For
large arrays this avoids memory issues.

Previously this returned True regardless of the value of
transientWindowType when really we want to also check if it is not None

- Default to 1 glitch
- Only try to calculate segment 2F if length > 1800s

 -new output: full (t0,tau) Fstat map,
  if bands in those params are searched
 -one file per Doppler grid point

Closes #7

Closes #9

 -replaces boolean 'transient' flag
 -window type [none,rect,exp]
  for both injections and searches
 -t0, tau bands for gridsearch,
  single-point use (e.g. MCMC) should be unchanged

Previously, a check was done to ensure that all initial positions of the
walkers would yield a non -np.inf prior. This extends this to also check
the likelihood evaluations. The method is also simplified partially

Partially resolves issues in #6

- Adds labels of each stage along the top
- Fixes bug when re-running
- Improves column title of run_setup table

Previously, we used the emcee.PTSampler, but this has been removed from
the master branch of emcee (see discussion here
https://github.com/dfm/emcee/issues/236) and a fork, ptemcee has been
developed. Testing shows this is equivalent (after some changes in the
interface) and perhaps better as it contains the adaptive temperature
ladders developed by Will Vousden (the maintainer of ptemcee)

 This removes the convergence testing ideas previously implemented
 (currently juts commented, but later to be fully removed). These are
clearly not useful without further study, which in itself would be a
better time to develop an implementation.