[DO NOT MERGE] make IndexRange agnostic of business logic

[sudeepdino008]

• if we're going towards "register domains/ii from anywhere", should kv.Domain also be string? -- I guess it's not needed now though. We can achieve the target of "make agg free of business logic" with kv.Domain as int as well.
• map feels ugly. can also just store the historyIdx (kv.InvertedIndex) in domain/history struct and iterate over to find the queried kv.InvertedIndex. But that seems inconsistent with what we did with aggregator.iis.

[sudeepdino008]

actually can also consider ditching map (for agg.iis as well) ...just using array and iterating over it when we need a "search" a kv.Invertedindex. Some functions that suffer are: aggregatorrotx.IndexRange, aggregatorrotx.IntegrityInvertedIndexAllValuesAreInRange; but they were already using switch statement before (which is sort of involves checking all cases). Thoughts?

[AskAlexSharov]

actually can also consider ditching map (for agg.iis as well) ...just using array and iterating over it when we need a "search" a kv.Invertedindex. Some functions that suffer are: aggregatorrotx.IndexRange, aggregatorrotx.IntegrityInvertedIndexAllValuesAreInRange; but they were already using switch statement before (which is sort of involves checking all cases). Thoughts?

I would prioritize performance of BeginFilesRO() func. Because it’s very hot and will be used in places which may not query any indices

[sudeepdino008]

closed in favor of #13459