Log duration on scrape failures (#54)

Currently we only log the duration of successful data collections but not on
failures. This makes it hard to understand the nature of eg. context deadline
exceeded.

This change moves the duration tracking up to make it availble for error logs
giving better insights into how long it took.

Author: Bjørn

main.go

@@ -287,8 +287,9 @@ func collect(ctx context.Context, client *client, organization org) ([]gaugeResu
 	for _, project := range projects.Projects {
 		start := time.Now()
 		issues, err := client.getIssues(organization.ID, project.ID)
+		duration := time.Since(start)
 		if err != nil {
-			log.Errorf("Failed to get issues for organization %s (%s) and project %s (%s): %v", organization.Name, organization.ID, project.Name, project.ID, err)
+			log.Errorf("Failed to get issues for organization %s (%s) and project %s (%s): duration %v:  %v", organization.Name, organization.ID, project.Name, project.ID, duration, err)
 			continue
 		}
 		results := aggregateIssues(issues.Issues)
@@ -298,7 +299,6 @@ func collect(ctx context.Context, client *client, organization org) ([]gaugeResu
 			results:      results,
 			isMonitored:  project.IsMonitored,
 		})
-		duration := time.Since(start)
 		log.Debugf("Collected data in %v for %s %s", duration, project.ID, project.Name)
 		// stop right away in case of the context being cancelled. This ensures that
 		// we don't wait for a complete collect run for all projects before