Martin/benchmarking

[Bouncner]

Some smaller changes to the benchmark script:

• randomize the merchant order and put some random sleep time in between
• discard stdout output of merchants
• start merchants before consumer is started (should be fairer in theory)

[Bouncner]

More or less driven by pain that our students like to pipe very large models to stdout. :) So I've either seen the "sold product" by the default merchants or vast model outputs.

I will make it optional and pass it as a parameter. I will make the output-suppression an opt-out.

[Bouncner]

Debugging output is no longer discarded by default.

Does this address all your concerns, @CarstenWalther ?