fix(decision): fix flaky TestTaggingUseful test

rather than use a ticker, use an actual timer that only resets after
sampling is completed. This doesn't affect behavior in real world
operations significantly but makes the tests much more reliable

internal/decision/scoreledger.go

@@ -134,8 +134,7 @@ type DefaultScoreLedger struct {
 // adjust it ±25% based on our debt ratio. Peers that have historically been
 // more useful to us than we are to them get the highest score.
 func (dsl *DefaultScoreLedger) scoreWorker() {
-	ticker := time.NewTicker(dsl.peerSampleInterval)
-	defer ticker.Stop()
+	timer := time.NewTimer(dsl.peerSampleInterval)
 
 	type update struct {
 		peer  peer.ID
@@ -149,8 +148,11 @@ func (dsl *DefaultScoreLedger) scoreWorker() {
 	for i := 0; ; i = (i + 1) % longTermRatio {
 		var now time.Time
 		select {
-		case now = <-ticker.C:
+		case now = <-timer.C:
 		case <-dsl.closing:
+			if !timer.Stop() {
+				<-timer.C
+			}
 			return
 		}
 
@@ -218,6 +220,7 @@ func (dsl *DefaultScoreLedger) scoreWorker() {
 		if dsl.sampleCh != nil {
 			dsl.sampleCh <- struct{}{}
 		}
+		timer.Reset(dsl.peerSampleInterval)
 	}
 }