Thanks for your replies!
MailGun already detects bounced emails at their end, suppressing those addresses in subsequent mail outs.
Is this analogous to the Discourse VERP process alluded to in that link ?
MailGun’s suppression of problematic email seems to be mandatory (they appear to want to protect their own reputation)
Would any purpose be served by trying to implement a VERP process within Discourse if there is no direct way for bounced email notifications to get past Mailgun and to reach Discourse?
I was wondering about another way to complete the bounced email feedback loop from MailGun back to Discourse; i.e. by passing on the suppressed address list that Mailgun generates, so that Discourse administrators, the users concerned and/or some sort of automated mitigation process are made aware and can attempt to resolve/restore delivery.
Am I making any sense? Do I have the wrong end of the stick?