Fix apparent data integrity violation with closed issues not actually being closed

Not sure whether this is the right tracker for this, but assuming https://ecosystem.atlassian.net is an OnDemand instance it seems appropriate at least:

I've received a few watch notification for AMKT issues just closed as Won't Fix by jkodumal ~4 hours ago, e.g. AMKT-4509/[AMKT-4510|https://ecosystem.atlassian.net/browse/AMKT-4510]/[AMKT-4602|https://ecosystem.atlassian.net/browse/AMKT-4602]/[AMKT-6358|https://ecosystem.atlassian.net/browse/AMKT-6358] - I've visited a few issues for details now and am slightly disturbed by what I've seen:

• they are still listed with Status=Open and Resolution=Unresolved
• likewise the last updated date seems to reflect the previous activity only
• yet all issue tabs properly reflect John's closing, i.e.Comments, History, Activity and Transitions Summary properly recorded him closing the issues as Won't Fix
• sure enough the issues still show up as Unresolved via JQL now, i.e. the current status is handled consistently at least, see e.g.

  project = "Atlassian Marketplace" and resolution = Unresolved and issuekey = AMKT-4602
  

• other than non formerly closed open issues they don't offer the Return to Triage transition anymore, so the workflow system seems to know about the intended current state Closed not offering that transition in fact

How is this possible? Am I missing something or is this a severe data integrity violation?