Calculation of credibility intervalls currently not via joint posterior

[shoepfl]

Hi there,

i recently figured out that the credibility intervals in pyPESTO are not calculated based on the x-percent HDI of the joint posterior but via percentiles for each parameter separately. This is for example in the function calculate_ci_mcmc_sample the case.

Wouldn´t it be better to calculate a posterior threshold based on an x-percent HDI on the joint posterior and afterward restrict the individual parameter CIs to the overall posterior threshold value? I think the other method could be problematic as individual calculation on each parameter does not necessarly have to match with the calculation on the joint posterior.

Best,

Sebastian