OK, I see what you mean.
I've done something similar in the past and rather than generating post-test artifacts using JMeter, I generated a URL link to the dashboard (Splunk in my case), filtered to the time period in question. Since I was pushing data from JMeter there anyway, and any investigation/interrogation would be done there, it made sense to just refer to the dashboard.
I think you'll have more success creating the backend listener to push JMeter data to Prometheus than configuring the perfmon plugin to pull data *from* Prometheus.
That way you'll have real-time JMeter activity being logged/displayed, rather than generating results post-test. And you'll be able to reference that directly, rather than pulling up a couple of JPGs.
To your question, I don't think there is any benefit of using JMeter to pull and store. Rather than JMeter being your source of truth, Prometheus (or Splunk. Or Sumo. Or ELK.) should be.