Rationalize command-line options

[hawkinsw]

Thanks for the prompt by @moeller0, we should rationalize all the command-line options so that they are consistent.

He asked the following questions to get the discussion started:

I have questions about these options:
a) why not add the units into the names? Say, --rpm.timeout_seconds 5 is less ambiguous than --rpm.timeout 5
b) why do we use different data types for both "percentage" measures?
c) --rpm.pct defaults to 0.05, is that really 0.05% or 5%, so ATM this seems to be rather --rpm.probe_traffic_fraction_of_capacity
d) --rpm.sdt is that in integer multiples of the standard deviation? If so, why only integer