The output is suitable for generating a timeline with data points for each time a task enters or leaves a column.

The column_metrics data set looks like this:

| project_phid | column_phid | task_id | ts_begin (unix ts)       | ts_end (unix ts)      | duration (seconds)     |
|--------------|-------------|---------|--------------------------|-----------------------|------------------------|
| PHID-PROJ-1  | PHID-PCOL-1 | 123456  | Time task entered column | Time task left column | How long task remained |

We can also derive another type of metrics from this same dataset - to calculate global workboard metrics and task-specific
metrics we simply summarize the above data.

* Each task will have a duration value for each on the workboard.
** the lead time for the task is the duration in default colum
** cycle time is total off all durations except for default column
* Global workboard metrics are the mean of all durations in each column.
** We can calculate the metrics using a moving window in order to capture the change in metric performance over time.

Change-Id: Ie416494100b4833006bd974a758ab27e4c04a6b4