#1 sse_cft1st doesn't work yet.

[Ben Herndon]

sse_cft1st inside of /client/opt/opt_sse.cpp currently
doesn't work as written.

The code is almost the same as sse_middleC, and calls
the same prep function. MiddleC works on 8 adjacent
tables of complex numbers with differing stride
sizes...basically cft1st is working with a table stride
of "one", where each complex is next to one another.

This makes it slightly harder to build the SSE 4 SIMD
registers, as you can't simply grab 4 at one time.

One way to solve it is to use two floats from the
current group and 2 from the next group to be
processed...basically process 2 groups at the same time.