Improve the bilby-mcmc processing tool

This makes a number of small changes to improve the behaviour and performance of bilby_pipe_process_mcmc:

1. Add a convergence-check option which creates a plotly HTML page enabling a deeper view of individual chains. It also calculates the simple Gelman-Rubin statistic and adds it to the title. Note: plotly is already a dependency of pesummary.
2. Enable the filter-match and filter-antimatch arguments to use regex matching
3. If only a single file is given and the --final flag is given, process it and produce a corresponding result file
4. Various minor fixes to ensure the created posterior is identical in format to a result file generated by run_sampler.
5. Add --no-conversion-function to enable quick-look tests
6. Add --output-directory to enable writing the output to a local directory (e.g. if processing results from another user)