replace interim-simplification 50 by simplify-each 30

[jberthold]

The --simplify-each option only affects booster execution while --interim-simplification forces a fall-back to kore which is more expensive.

[jberthold]

I have tested this locally with a P-token proof (test_ptoken_domain_data) and saw around 17% speedup (6'42'' to 5'33'')
Removing the simplify-each causes memory issues in the LLVM backend and huge unevaluated terms on the <locals> cell and anywhere the locals are used in computations (there are a few functions that do, including getLocal).

[Stevengre]

Thank you! LGTM!