some analysis:

 
this job had 4 red dots in 30 runs over the course of a month. Only one 1 red dot was an infra abort.

this job had 13 runs over a month, 2 of which were red dots. One 1 of those red dots was an infra issue.

this job had 30 runs over a four day period with two red dots. Neither red dot was a problem with infra.

this job also with 30 runs over a four day period had 3 red dots, none of which were infra related.

To summarize, that's aprox 120 jobs with 11 red dots, but only 2 were infra related. So, essentially we have a 1.6% rate of infra
aborts. Also, the 2 infra aborts we saw were not some indication of infra being broken and affecting jobs running afterward.

I'm not sure it's justifiable to spend time on automation to warn the right people if we get an abort, unless someone has
the extra cycles or there is some other problem I'm missing. This is not to say we didn't have major issues in the past when
something like this would have been very useful.