Apologies for being so dismissive of your point above. Both my tone and my point were wrong. Let me try again:
Of course, you are right. Changes in the make-up of the field (what statisticians call "composition" effects) matter. And, as you point out, the composition of the field at the Hemery meet has changed dramatically in the last few years.
A quick look at this year's results shows that I was wrong to dismiss your point" ~17 of this year's sub 4 miles were run by pros; last year, 1 was. (I say ~ 17 b/c I counted the "unattached" as pros.)
That leaves 35 college kids running under 4. To your point, many of many of them probably weren't running this meet in 2016. Figuring out who would and wouldn't have will take some time. Until then, my quick look shows a lot (~10 college kids from the NE running under 4 this year).
Obviously, some of them got dragged along to faster times by pros or faster college kids.
The beauty of looking at times at BU is that we don't have to worry about the track being faster. But, as you point out, we still have to worry about the fields getting faster.
Again, my apologies for my earlier tone. Your point was a fair one -- even a quick scan shows that taking it into account makes the headline less impressive.
At the same time, it sure looks to me as if the bump isn't simply a composition effect. But it will take a bit more time to see if that hunch is correct.