Coloc version 1.10 on CRAN. This version adds the option to calculate a credible interval for eta from its posterior distribution.
It uses a rather crude search to find the highest posterior density region, based on a numerical approximation of the posterior. The search proceeds outwards from the mode in largish steps until the desired level of the interval is exceeded, then inwards in small steps. It could certainly be optimized, but this way was quick to program and runs in a few seconds, so appeared reasonable. The calculation can be turned off by setting
bayes.ci=FALSE to speed things up when time really matters in simulations.