Test apps s_time with new -testmode option

[bukka]

This tests apps s_time that is currently completely untested.

It is composed of 4 tests and runs around 8s in total (not sure how to make it quicker as it's using min time). The s_time is currently not tested at all so it will improve the apps coverage a bit.

EDIT: A new -testmode param was added so the test takes only 800 ms which should be acceptable.

[bob-beck]

s_time is really just a connection counter and averager, while ok, sure, it improves test coverage I'm not sure I want 8 seconds of extra time expended on every time I run the tests to verify this. I run tests a lot.

[paulidale]

The time this takes will be in parallel to other tasks and shouldn't increase the overall run time.
Except on single threaded test runs which ought to be uncommon.

[bukka]

Yeah it will run in parallel so real slow down won't be that big.

The main reason for the test is that the tool is not going to regress easily and testing multiple session reuse might be useful as well I guess - think it is already tested but usually just with few reconnections. I would need to check that part if it brings anything new though. Still having some tools with 0 coverage is not ideal (although not sure if it's worth it to spend time on things like srp just to increase coverage).

[t8m]

We have special-cased similar test use case in openssl speed by adding an option. We could for example special-case -time 0 as do just a couple iterations. This is for master branch only anyway.