Say once after a restart that nothing waits, so an old retirement condition clears (issue 937)

The controller clears a provider's retire-waiting condition only on its
settled word, and a provider says settled only for a set its own memory
held. A provider restarted after the set stopped waiting (here: the
delivery of the grant-file fix) never says it, so the condition kept
asking the operator to approve retiring what nobody waits on. The first
pass of a process that finds every consumer asked now says settled once,
with no set; the controller clears both conditions on it.

modules/postgres/cmd/postgres-provider/retirement.go

@@ -105,6 +105,9 @@ type retirement struct {
 	lastWant  map[string]bool
 	seeded    bool
 	seedSaid  string
+	// noneSaid is whether this process has said that nothing waits: an earlier process may have left a set
+	// waiting that this one never knew (issue 937).
+	noneSaid bool
 }
 
 type waitingSet struct {
@@ -257,7 +260,15 @@ func (h *Harness) settle(why string) {
 		set = h.r.waiting.set
 	case h.r.rejected != nil:
 		set = h.r.rejected.set
+	case h.r.noneSaid:
+		return
 	default:
+		// A restart forgets a set an earlier process waited on, and so would never say it ended: the
+		// controller's condition would ask a person to approve what nobody waits on any more (issue 937).
+		// The first time this process finds every consumer held asked for, it says so once, with no set.
+		h.r.noneSaid = true
+		h.say("nothing waits for a person: %s", why)
+		h.announce(EventRetirement, h.retirementBody(ChangeSettled, []map[string]any{}, map[string]any{"why": why}))
 		return
 	}
 	h.r.waiting, h.r.rejected = nil, nil