Simplemma analyzer

[osma]

Just like with spaCy (see #374) we could add an analyzer that uses simplemma for lemmatization. This is a very fast and lightweight multilingual lemmatizer which currently supports 38 languages. The lemmatization accuracy may not be as high as with e.g. Stanza (see #539) but in practice that doesn't seem to matter much based on experiments I've performed.

Simplemma is implemented in pure Python without external dependencies so I think it should be possible to include this as a core feature, not an optional one, unless there are any unexpected problems with e.g. supported Python versions.