jackknife pair counting weights

[duncandc]

I cannot find the citation to justify our jackknife pair counting weighting scheme.

Currently we count pairs as:

1. both within the removed sample as 0.0
2. one within and one outside as 0.5
3. both outside as 1.0

The straight forward thing to do would be to count (2) as 0.0. Thoughts @aphearin @manodeep ?

[aphearin]

@andrew-zentner - do you know of a reference for this?

@duncandc - have you tested how much of a difference this choice makes for a standard HOD mock, or just a halo sample at fixed mass?

[manodeep]

I can see the reasoning behind using 0.5. May be we can design some test cases to figure out the correct form.