Bug: Cannot use columns with `RequiresOptIn` types in Notebook

In notebooks, whenever an accessor is generated for an experimental type that requires `@OptIn`, we get compiler errors:

This makes sense, as each usage of that type, or the entire snippet, would need to be annotated with `@OptIn`, however, that does make it currently impossible to store any experimental type in a DataFrame and use it.

We could try to ask the notebook team what to do about it, or simply introduce a list of opt-in classes for `dataFrameConfig` in notebooks.

Start by reproducing the notebook compiler errors when a DataFrame contains a type requiring @OptIn, then inspect the notebook dataFrameConfig entry point. Clarify with the notebook team whether notebook-wide opt-ins or a configured list of opt-in classes is expected; done means experimental types can be stored in and accessed from DataFrames without compiler errors.

Written by the indexing model from the issue text.