Parallelize locator lookups

[hannahhoward]

Goals

Enable usage of the Indexer's multihash lookup

Implementation

• Extract common logic in batching to its own class
• Modify #executeReadClaims to call with multiple hashes
• Modify #internalReadClaims to put requests in a queue if unless explicitly told to be synchronous

For discussion

I'd like some feedback before I finish testing and fix some of the tracing. anecdotally, this seems a win in testing with the indexer.

[fforbeck]

It looks good to me. Added one minor suggestion. Two questions, not necessarily affected by this change, but still related 1) is it ok to clone the blob before returning it in the resolveRequests? 2) if the fetch blob fail in processBatch, then it stops the flow, should we log the error before the iteration break?

[fforbeck]

Suggestion: Expose as a configurable parameter if batch sizes need to vary based on CF worker limitations.