Fix data race on metricDescriptorsFunction start and end times

[thepalbi]

Doing some trials with the exporter, I found a data race condition in the go-routines launched by each metricDescriptorsFunction when about to pull time series.

The raw stacktrace goes as follows:

==================
WARNING: DATA RACE
Write at 0x00c00000f7b8 by goroutine 104:
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func1.1()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:245 +0x834
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func1.2()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:283 +0x58

Previous read at 0x00c00000f7b8 by goroutine 41:
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func1.1()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:260 +0x1240
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func1.2()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:283 +0x58

Goroutine 104 (running) created at:
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func1()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:223 +0x3c8
level=debug ts=2022-05-19T15:42:57.467941Z revision=fd8d54ee-WIP caller=monitoring_collector.go:255 traceID=1272ddad166d68b1 instance_id=123 metrics_id=123 job_name=holis msg="retrieving Google Stackdriver Monitoring metrics with filter" filter="project=\"wired-height-350515\" AND metric.type=\"serviceruntime.googleapis.com/api/request_latencies_backend\""
  google.golang.org/api/monitoring/v3.(*ProjectsMetricDescriptorsListCall).Pages()
      /Users/pablo/go/pkg/mod/google.golang.org/api@v0.75.0/monitoring/v3/monitoring-gen.go:10545 +0x180
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func2()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:311 +0x564
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func3()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:314 +0x58

Goroutine 41 (running) created at:
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func1()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:223 +0x3c8
  google.golang.org/api/monitoring/v3.(*ProjectsMetricDescriptorsListCall).Pages()
      /Users/pablo/go/pkg/mod/google.golang.org/api@v0.75.0/monitoring/v3/monitoring-gen.go:10545 +0x180
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func2()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:311 +0x564
  github.com/prometheus-community/stackdriver_exporter/collectors.(*MonitoringCollector).reportMonitoringMetrics.func3()
      /Users/pablo/work/stackdriver_exporter/collectors/monitoring_collector.go:314 +0x58

Diving a bit deeper, the data race arises when one go routine, launched here, tries to write the endTime object; while another one attempts to read endTime. Since the endTime date object is shared across all go-routines this leads to a data race.
This is easily fixed by passing both startTime and endTime by copy to each MetricsDescriptor processor go-routine.

Notice that this problem ONLY happens if the ingest delay flag is turned on, because the offending write path is flagged with it.

[SuperQ]

Ohh, good catch.

This needs a DCO sign-off. You can use git commit -s --amend to add it.