SV term in LSSWeights Covariance

[elvinpoole]

The 2-point functions computed in the Density trend SV covariance are likely overkill (too many bins, too low bin_slop, etc). we should run some realistic tests (e.g. on the DES Y1 example) and pick some more sensible defaults