Use one of MH resampling or Stirling approximation in auxiliary beta variable resampling

[jamesross2]

Currently accomodates both approaches, but this customisability
is more confusing than helpful. To decide on better approach, need
to test both

• speed of resampling under each approach, and
• speed of convergence for chain.

Most likely will go with MH resampling, since this is better understood by most users.