Speed up ingestion: get rid of TimeHistogram #12615
Description
When ingesting high-frequency, low-bandwidth data (e.g. scalars) we spend a significant amount of time updating the TimesPerTimeline data structure.
We currently use TimeHistogram for two things:
- A) go to next/previous time step
- B) Gap detection for timeline
A) we should be able to do with a much simpler approach (e.g. just using ChunkStore).
B) is a bit more difficult, but we should be able to accomplish something that is much faster, somehow.