Integration of extensions

With PEtab v2 we will introduce extensions to PEtab that will add new functionality. (In exceptional cases, extensions might also change some existing rules, but this is for another issue). For example, there is petab-select for model selection, units support for PEtab, and there were discussions to move the petab v1 visualization functionality to an extension.

Ideally, those extensions would be able to extend the current PEtab validator. Possibly they would also want to extend the petab.Problem API. Not sure what else.

The question is, whether those extensions should be directly incorporated here in libpetab, or whether we provide some plugin mechanism (via pluggy?). Currently, I'd prefer the latter, to avoid bloating this library and allow for mostly independent development of the extensions. However, this might make it difficult to handle any interaction of extensions. Happy to discuss other options and relevant points for plugins to modify libpetab behaviour.

Start by reading PEtab v2 extensions in PR #537, then review the current PEtab validator and petab.Problem API mentioned in the issue. Compare direct incorporation with a pluggy-based mechanism, including how extensions might interact. Done means documenting and implementing an agreed extension integration design.