Will take a look at this one but am inclined to reject it.  The timing 
benefits are marginal at best and only help in the atypical case of 
mixed floats and ints. The patch changes the type and ordering of 
intermediate sums, resulting in different answers than the standard sum
().

In contrast, the existing fast sum() was carefully designed to mimick 
the operations of the original sum(), so it always produces the same 
results.