The issue is connected to dop251/goja#190 in the
sense that instead of the float to int conversation to result in either
error or a clamped float value, it results in a 0 int value which panics
the stdlib time package.

Changing the parameter for the timeout/interval to float both makes it
possible to take values such as 0.5, meaning half a ms, and also checks
if the value is positive so we can be certain that it won't panic k6.
If not this results in an exception in the script.

Co-authored-by: Ivan Mirić <ivan@loadimpact.com>