Create separate trend dashboard, graphing key metrics over time.

[temyers]

This PR creates a CloudWatch dashboard for graphing pipeline metric trends over time.

This pipeline-dashboard provides great tooling for assessing your project health. However, without trend information, you cannot determine whether you are improving.

The pipeline dashboard includes the following trends:

• Deployment Frequency - aligns with the DORA State of DevOps report metric of the same name.
  • Annottations are included for "Elite" and "high" performing thresholds
• Lead Time - Graphs the same lead time metric as built into pipeline-dashboard - also aligns with the DORA metric.
  • Annottations are included for "Elite" thresholds
• MTTR - same as existing pipeline-dashboard
• MTBF - same as existing pipeline-dashboard

Current Limitations:

• The time units are not configurable - each dashboard is configured with set units, converting from seconds. The choice has been made based on DORA metric recommendations. If a user has really poor metrics, it could be a nudge for changing behaviour. "If something hurts, do it more"!
• Due to CloudWatch limits, a maximum of ~30 pipelines can be displayed on a single dashboard. This should be sufficient as an initial release - I'd consider > 30 pipelines in an account as a potential smell - using a multi-account strategy, fewer pipelines should be expected.
  • In the event of too many pipelines, a warning shall be logged. In practice, without an alarm this is likely to be ignored, but is provided for information.