This milestone is related to merging our new bayesflow interface build on Keras 3 (currently living on the dev branch) into main, thus officially releasing what is essentially Bayesflow 2.0 given that the codebase has been completely refactored.