MFC r343924:

Correct the CPU target for powerpcspe

The MPC8540 is actually e500v1, which doesn't have double-precision floating
point support. The 8548 does, so use that as the CPU target.