Calculate a unique ID for aggregations

This will allow for smart diffing on the data layer. The ID will need to be a serialization of the following information:
- the order of the events in the aggregation
- the ID of each of those events
- the duration of each of those events (maybe?)

This will allow a smart diffing at the data layer as aggregations evolve over time. For instance, using Backbone:

`aggregations.set(newAggregations);`

will intelligently add and remove aggregations from the collection.

Start by reading the aggregation implementation and its current output shape, then trace how Backbone's aggregations.set(newAggregations) consumes IDs. Decide what event order, event IDs, and optional durations must contribute to a deterministic serialization; done means evolving aggregations can be smart-diffed without treating unchanged aggregations as new.

Written by the indexing model from the issue text.