[WIP] feat(core): Experimental support for out of order samples (POC)

[vishramachandran]

Done:

• seq Number label placeholder
• creation/find new time series with increased seq number when OOO is encountered
• query works

Open Issues

There is an edge case where there is a very small OOO ingestion possibility on shard restart. From a purist perspective, here are issues to be addressed

• (Doable) Lucene index needs update with new startTime
• (Doable) Lucene index needs to retain larger endTime (which may have occurred much before) when TS stops ingesting
• (Doable) Chunked & Sliding Window iterators need to account for overlapping chunks
• (Hard) We cannot always be sure all ingested chunks are in memory given possibility of overlapping chunks in history. Should we ODP each time, for every query? This is a hard one