sg/msp: support for super-simple alerts on custom metrics

[jac]

Closes CORE-16

Currently service owners needs to either rely on our default alerting or use terraform to add custom alerts. With this PR service owners can specify simple custom alerts using MQL or PromQL in the service file e.g.
(ignore bad alerts; just for demonstration purposes)

monitoring:
  alerts:
    customAlerts:
      - name: Billable Time MQL
        severityLevel: WARNING
        description: "Check if mean billable time is below 30 seconds"
        condition:
          query: >
            fetch cloud_run_revision
            | metric 'run.googleapis.com/container/billable_instance_time'
            | filter resource.project_id == 'msp-testbed-test-77589aae45d0'
            | group_by 30m,
                [value_billable_instance_time_mean: mean(value.billable_instance_time)]
            | every 30m
            | group_by [resource.revision_name],
                [value_billable_instance_time_mean_min:
                  min(value_billable_instance_time_mean)]
            | condition val() < 30 's'
          type: mql
      - name: CloudSQL Bytes Used over 1000000000
        severityLevel: CRITICAL
        description: a lotta bytes
        condition:
          type: promql
          query: sum(avg_over_time(cloudsql_googleapis_com:database_disk_bytes_used{monitored_resource="cloudsql_database"}[30m])) > 100000000
          durationMinutes: 2

Test plan

generated alerts using service file config and viewed them in msp alert dashboard

[unknwon]

🚀

[bobheadxi]

Nice 😁 Thank you!