hwclock/systohc seems to be flaky on Travis CI

[evverx]

To judge from https://travis-ci.org/github/karelzak/util-linux/jobs/703166322, it seems the test fails from time to time with

      hwclock: system to hw                   ... FAILED (offsets 216.218.254.202: +0.018064 -0.998979 +/-1.017043)

[karelzak]

Hmm... frankly use hwlock in this environment is probably fragile. I guess it's some container or virtual machine and in this case do anything with HW clock or assume small latencies is a bad idea.

Maybe we can detect that at the beginning of the test that it's virtual machine or container and call ts_skip rather than execute the test (or introduce ts_skip_nonhw function).

[evverx]

I run that test in VMs and containers and it seems to be fine there so I'd probably skip it on Travis CI only for the time being.

Looking at https://travis-ci.org/github/karelzak/util-linux/jobs/704172347 where it's just failed, it seems it happens on Travis CI when dist is precise so probably I could call ts_skip when those conditions are met. I need to figure out what environment variables I should check though.

[evverx]

It should be addressed in #1090