Allow storage compaction

[FlorianLautenschlager]

Currently the API is designed to handle hole time series objects at once.
But there is also the use case to store single time series data points. This will of course lead to a vast of lucene documents and lower the compression rate and query performance.
For that one need a storage compaction that collects, groups and reduces single lucene documents into one "large" document.