The closest we have currently is probably Runtime warnings and convergence problems (with work on an update ongoing). Unfortunately my impression from answering the questions here on Discourse is that there actually are very few “typical errors” and most problems are quite unique or at least require a lot of background knowledge across stats, programming and numerical math to let you see how they connect to other problems. So while I think we can somewhat improve on the “typical errors” / cheatsheet /… front, I think a bigger, more important and harder task is to find ways to make the necessary background knowledge more accessible and discoverable.
In any case, users sharing their own problems and solutions here or anywhere else (e.g. personal blogs) IMHO has a side effect of letting other users learn, so that’s great.