Class CheckHits

Some explains methods calculate their values though a slightly different order of operations from the actual scoring method ... this allows for a small amount of relative variation

In general we use a relative epsilon, but some tests do crazy things like boost documents with 0, creating tiny tiny scores where the relative difference is large but the absolute difference is tiny. we ensure the the epsilon is always at least this big.

Tests that a Hits has an expected order of documents

Tests that two queries have an expected order of documents, and that the two queries have the same score values.

Tests that all documents up to maxDoc which are not in the expected result set, have an explanation which indicates that the document does not match

returns a reasonable epsilon for comparing two floats, where minor differences are acceptable such as score vs. explain

Assert that an explanation has the expected score, and optionally that its sub-details max/sum/factor match to that score.