In the recent updates to the dynesty api to support dynesty 3, both the bound and proposals options have no effect when selecting any of the bilby-implemented samplers. As a result, only differential evolution can be used as a proposal scheme, and the volumetric proposals would be very inefficient as they would always use the identity matrix as their proposal axes.
For the proposal method, there is a one-line fix in dynesty.samler_init_kwargs.
For the bounding, it is a bit more subtle because with dynesty3, all dynesty bounding options should just work for the volumetric proposals.
For differential evolution, they would have no effect anyway. So it would be ok to completely remove any reference to the 'live' bounding and rephrase the documentation a bit.
I have implemented both these based on #1049 dropping dynesty<3 support here.