RE: Composites The problem is that given some (possibly a lot) of the historic holes predate 1980 or even 1990, there may be no way to "rehabilitate" or verify the data from those holes. At a minimum you would need original assay certificates, good drill hole location and orientation data, and preferably the ability to re-run some of the historic samples. In my experience with holes of this sort of age, this is extremely difficult to do. Often such old holes really only help to serve by outlining areas to redrill using modern techniques and proper QAQC protocol. The widespread use of good QAQC (standards, blanks, duplicates) really only dates back to the Bre-Ex era in the early 1990's. Most work before that will not be NI43-101 compliant.
Also, work from earlier resource calculations on Cow Mountain (Giroux) has already shown that a large number of historic holes were not suitable for resource work due to some of the above reasons and were excluded then. I see no reason why they would all of a sudden become good at a later date.
Your analysis of how many composites exceed 5 g/t Au is a good one and should illustrate to people exactly why there is so much scepticism over the 10Moz number, whether is be due to lack of cutting, improper compositing, overly generous 3D solids that cover areas of no assay (-1's), and potential smearing of high grades.