Optimize Performance of Timeseries Integration Layer

The Timeseries Integration Layer is feature complete. However in order to use it in production it must be performant enough.

Opennms installations get around 40.000 datapoints (Metric + timestamp + value) per second.

This task is to examine the bottlenecks of the integration layer and to remove them . The current implementation is based on the Opennms Newts code and optimized for Newts. We can now take this the next step and optimize for the integration layer needs.