Change how host timeouts are initialized in ultrascan.

[jsiembida]

When targets are subject to congestion control, their timeout clocks
should not be initialized to the same time, otherwise they are being
completed without even trying. Command that triggers the issue in AWS:

nmap -sT -T3 --host-timeout 5 -n --max-hostgroup 100 -Pn -p 8080 a.b.c.d/24

[dmiller-nmap]

This is an interesting idea. I can see how the existing behavior could be a problem. I will have to do more review and testing to see if this is a complete fix, but for now, here are my concerns:

1. What happens to the startTimeOutClocks function? Is it called anywhere else? Could it be eliminated?
2. Why call gettimeofday in the debug statement, when we are actually using a cached timeval (USI->now) to set the timeout clock? The debug statement should print the actual time used, or it should be eliminated.
3. Does this problem affect the other startTimeOutClocks functions in service_scan.cc and osscan2.cc? Should those places be changed, too?

[jsiembida]

1. Left "just in case", but it is not used any more.
2. Overlooked that.
3. I am not sure, only fixed the scanner we are using. But a quick look suggests the timeout logic is similar, therefore likely to show the same behavior.

[dmiller-nmap]

I've made slight adjustments and will be committing this soon. Would you like a different name or full name credit in the CHANGELOG, or just your Github username?

[jsiembida]

I am happy with whatever is more convenient to you.