Restartable chase with changed program

[mmarx]

We should cache the results of SPARQL queries as part of the session state.

We should be able to reason over a program, change some facts or rules, and then continue reasoning with the changed program, while keeping all unaffected inferences. In particular, this would be useful for caching the results of SPARQL imports when using nemo-web.

[mkroetzsch]

This might have become quite a bit more complex with the advancements on how Nemo handles SPARQL. I was wondering if this should maybe become a backend functionality. The idea is as follows: when the program is changed and Nemo is run again, then let Nemo try to re-use as much of its previous computation as possible. This would involve previously loaded inputs, but it could also involve whole strata of computation (e.g., if only some detail in the final output rules changed). With this view, one could preserve auxiliary binding relations and partial SPARQL imports as well.

[monsterkrampe]

Yes, I think we cannot handle this in Nemo Web anymore since all web requests are issued directly from WASM now. So I'm moving this issue to the Nemo repo.

[mmarx]

It seems that having a way to restart the chase with some changes to the program, while keeping all relevant intermediate values, is indeed the way to go here.