iperf tests handle under-speed links poorly

When the ethernet/multi_iperf3_nic_device* tests detect an under-speed link (say, a 10 Gbps device hooked up to a 1 Gbps switch), the result is a rather awkward and lengthy batch of output, like this:

INFO:root:-------------------- Test Run Number 1 --------------------
ERROR:root:Detected link speed (1000) is lower than detected max speed (10000)
ERROR:root:Check your device configuration and try again
INFO:root:
INFO:root:-------------------- Test Run Number 1 --------------------
ERROR:root:Detected link speed (1000) is lower than detected max speed (10000)
ERROR:root:Check your device configuration and try again
INFO:root:
INFO:root:-------------------- Test Run Number 1 --------------------
ERROR:root:Detected link speed (1000) is lower than detected max speed (10000)
ERROR:root:Check your device configuration and try again

This repeating error goes on for quite a while, which both pointlessly extends test time and creates a very verbose attachment that provides very little real information. (This example is taken from https://certification.canonical.com/hardware/201309-14182/submission/119573/test/64027/result/8817033/.) Two possible better ways to handle this would be:

* Stop the test as soon as an under-speed link is detected. After
  all, it's not likely to get better while the test is running.
* Go ahead and run the test despite the under-speed link. The
  test is virtually certain to fail, but at least we'll get
  some data. This might be acceptable on, say, a 40 Gbps link
  detected when a 56 Gbps link is theoretically possible; we
  could attach a note saying that the device was tested at
  40 Gbps rather than the theoretical maximum of 56 Gbps.