results of intense empirical testing

- calculation of mbit/s was buggy
- after a lot of tests the basic (unweighted) calculation was the most precise
- only use large packtes for testing, since small packets have overhead wich is dramatically large in relation to package size

- closes https://github.com/iqb-berlin/testcenter-iqb-ng/issues/17
- closes https://github.com/iqb-berlin/testcenter-iqb-ng/issues/16