limit size of justification for which we do forward reasoning

[tobiaswjohn]

e.g. benchmark pipes
very long justifications --> initialization takes a lot of effort, but gain is very small --> increases overall time

solution (not pretty but should work): add upper limit on number of variables (e.g. 5?, 10?) until which we calculate the schema