To avoid [strange failures](https://github.com/oda-hub/nb2workflow/actions/runs/9143135679/job/25139444344) At least sparql queries seems to be not thread-safe in rdflib And also think of better optimisation than the whole graph being global variable Also, don't duplicate restrictions (blanc nodes) when parsing annotations
To avoid strange failures
At least sparql queries seems to be not thread-safe in rdflib
And also think of better optimisation than the whole graph being global variable
Also, don't duplicate restrictions (blanc nodes) when parsing annotations