- rework how the Pymc3 sampler class sets the priors

 - move the setting of the PyMC3 priors into the Pymc3 sampler
   class

 - this is the start of adding functionality to use the PyMC3
   sampler. There is a lot more that needs doing, but the basics
   work.

Allows users to give priors as an OrderedDict

See merge request Monash/tupak!142

By putting in the dictionary as an ordered dict, the way priors are
entered determines the order they are plotted, which is what most people
will expect.

- Also adds some catches to help improve error messages for when the
  input prior is something else.

Fixed missing function parameter in several function calls for `asd_from_freq_series`

See merge request Monash/tupak!141

Resolve "`plot_data` in `Interferometer` is not correctly normalised"

Closes #156

See merge request Monash/tupak!140

Adds more example tests to the CI

See merge request Monash/tupak!128

Simplify wg redundant code

See merge request Monash/tupak!124

Add Poisson likelihood to core likelihood functions

See merge request Monash/tupak!132

Updated the eccentric inspiral example

See merge request Monash/tupak!137

 - complete the radioactive decay example to test the Poisson
   likelihood
 - refs Monash/tupak!132