Duplicated log output in prior/posterior predictive sampling

[ricardoV94]

I get this issue often after the changes in #6142 in jupyter environments.

If someone can reproduce and provide a minimum example that would be great.

If someone knows how to fix the problem, even better!