Posixlib times method needs more validation

[LeeTibbert]

PR #3087 modifies TimesTest so that it does not fail CI when times() returns a zero
user or system process time. I can not find any obvious error in the implementation
of posixlib times() but I do not trust it. I believe that it needs more validation.

That validation probably needs to be done manually, so that it does not cause
CI to intermittently fail. It probably needs to be in a sandbox project in order
to have better visibility of the actual reported user cpu times and better
control/expectation of what 'reasonable' values should be. Right now,
they are tested for 1) changed & 2) non-negative. I suspect the reported
values are way smaller than one would expect. No proof, though.

I am capturing this experience in an Issue in case others use times() and
get weird/wrong values.

After a good manual Test or Tests are created, perhaps they could be
folded into CI. Or perhaps they need to remain manual.

A good project to peck away at whilst waiting for other PRs to clear CI.