It could happen in rare occasions that the subprocess survived death of
the tests, mostly when exceptions happened in the parent process and it
didn't kill it. So there's a finally now and the subprocess has a
limited life length just in case it still slipped somehow without kill.

The tests only received data, but did not check it is correct. Now it is
checked.

Using boolean parameter instead of cryptic flag-like one. It is less
flexible, but the flexibility isn't really needed.

Pointed out in review.
A test that would catch it needs to be written.

At last longer than is the dead-socket timer, to see the socket is not
considered dead even when it is read.

Also, fix a small bug that could lead to resource leak (but probably
wouldn't)

* Added a ping command to msgq, for testing purposes
* Sending large and small data around from the test

* Tests that try sending data and expect to time out.
* Test that tries to repeatedly send a command to msgq and expects it to
  be answered (to see that the sending does not only timeout, but works
  when the data are read as well).

This reverts commit a669db43.

The reverted commit was introduced due to misoperation possibly with
git-new-workdir, and actually reverted previous changes made on master.
By reverting it we now recover these lost changes.

Conflicts:
	ChangeLog

it involves refactoring of the AuthSrv class: It now internally constructs
IOService; the timer is also constructed internally, using the IOService
object.

the timer.

other tests.

and removed an unused variable "PROGRAM"

Most obvious change is for the command "Stats show", where the blob of stats data is now printed ina more readable fashion.
Also removed the print of 'send the command' line, and empty results are not printed at all