During a nested sampling run, you may wish to update the internal parameters
of the LRPS based on sampler performance or other criteria. The frequency of
these updates is set by the first_update and update_interval arguments of
ernest_sampler().
If you are creating your own ernest_lrps subclass, implement this method to specify any special update behaviour. The default method reconstructs the LRPS with current parameters and resets the likelihood call counter in the cache.