feat(Searcher): Ignore old data from long running filter responses

Currently, for filter results that take a long time to resolve, if the user types in filter text slowly, then the results that come back are not in chronological order, leading to incorrect results that do not match the current filter text.

For example, the user types:

b => query A bu => query B bull => query C

But the responses come back as:

response C => 5 results response B => 500 results response A => 5000 results

And so what the user sees on their screen is an immediate 5 results, followed by more and more results that don't match the current query.

This introduces an optional timestamp that will ignore previous results that were initiated prior to the latest.

See video for repro.

https://user-images.githubusercontent.com/3843503/166277915-fe7c8e52-8876-4b91-a062-200fd67cfd71.mov

bullhorn / taurus

feat(Searcher): Ignore old data from long running filter responses #81