fix gh-2065

Mixed

This was discovered by the daily buildbot.

Resolved issue #2055

Doc

Reflow paragraphs
Remove flops() since it's not really important.

Faster algorithms and gradients for GpuCorrMM

Struct support

This breaks all ops that use it otherwise.

and clarify/update documentation.

bugfix in configparser

Remove the `double round(double)` in pymath.h for Win32 only by defining
`HAVE_ROUND` via the `-D` switch of `nvcc`.
The `double round(double x)` function is present in CUDA
6.5.13/6.0.37/5.5.22/5.5.11/4.1.28/4.0.17, so I think it's safe to remove
the one in `pymath.h`.