Set default dynesty sampler sampling method to auto

Rather than using the default dynesty sampling method applied via bilby, which is rwalk, use the auto option. For the standard 4-parameter problems using PE this will cause dynesty to default to uniform sampling, which is more efficient and faster than rwalk in our cases. If the user want to use different settings then they can specify them in their PE job configuration.

In fact, when using one of the documentation examples, uniform sampling starts well, but then takes ages. I could potentially play around with the bounding options to try and find a better set, but using rslice seems to be quick and work well, so I'll default to that.

Closes #71 (closed).