BandwidthTestHandler: avoid Gitlab messages when builds have to be restarted
lhcb/Rec!3664 (comment 7940676) recently showed that, in the case of the builds having to be restarted, there are a couple of our Gitlab messages from the BandwidthTestHandler that will be printed and cause confusion. This MR protects against by only allowing messages through if the handler is running on the "test" build - we'll then only get messages to the MR corresponding to that test build.
I know you've told me to watch out for this before @shunan - it seems I didn't catch all the edge cases until now.
Have confirmed on a local test that the messages should go away.