SharpeRatio.haircut is very slow for large audit environment

The SharpeRatio.haircut function is very slow for a large audit environment.

It seems likely that loops and especially memory allocation inside those loops is the culprit. There is likely also functionality that can be vectorized.

Locate the SharpeRatio.haircut function in the R source and profile it with a large audit environment to identify loop and allocation costs. Compare any optimization against the existing function's results; done means materially improved performance while preserving equivalent output.

Written by the indexing model from the issue text.