Add "aggregate" and "rolling_up" features

This is an umbrella blueprint for such features as:
 - Aggregation data
 - Queries precalculations (later)
 - Rolling up sample data
All of these features have one thing in common: there should be some 'manager' that will start corresponding process (e.g. 'aggregation' or 'archivation'). Besides, I believe that the same logic and even the code may be reused for implementing all the features. Aggregation is the same as rolling up but without data removing and so on. So all of these features should be very flexible and we need to work out the pluggable architecture that may be easy extensible.