implement efficient set intersection + iteration for RLE bitsets

[lukevalenty]

No description provided.

[AshleyRoll]

I'll start tackling this issue next.

Plan is to build a variadic template that will take N rle_indexs and walk though the RLE data and apply bit-wise and operation. This means that we should be able to:

• Take the maximal length of 0 bits from the N RLE streams
• If no 0s are present, we take the minimal length of 1s
• Then increment each of the RLE stream states by the decided number of bits and repeat

This should then give us a fairly efficient way to combine the RLE data streams into an overall intersection. And then I should be able to turn that into an efficient iteration for running callbacks.