Removes opencv dependency

[fran6co]

Rebased and squashed #261

[floe]

@larshg OK if I merge this one instead of your original PR? Your contributions are still part of the commit log, of course.

[fran6co]

I squashed some changes into larshg's commit for standardizing the timer and the inclusion of the opencv header.

[fran6co]

Wouldn't it be better to remove the opencv dependency for good?

[larshg]

@floe sure, as I don't have much time right now it's nice someone continues the work :-)

[fran6co]

Removed opencv dependency, the only problem would be test_opengl_depth_packet_processor.cpp. But I think it needs it's own cmake rules

[floe]

Could you add division by 4500.0f in the grayfragment shader, then I think we're good to go.

[fran6co]

I don't have the kinect at hand at the moment to test it out. Could you check if it's okay? https://github.com/OpenKinect/libfreenect2/pull/361/files#diff-8ca25d621f26a3215be9537b77208b46R59

[xlz]

The last commit introduces hard dependency on C++11. As discussed before in #261, we want to avoid this.

Can you add a wrapper like this?

double getTime()
{
#if __cplusplus >= 201103L
  //not tested:
  static auto start = std::chrono::high_resolution_clock::now();
  auto end = std::chrono::high_resolution_clock::now();
  auto duration = end - start;
  return std::chrono::duration_cast<std::chrono::duration<double>>(duration).count();
#elif LIBFREENECT2_WITH_OPENGL_SUPPORT
  return glfwGetTime();
#else
  return 0;
#endif
}

I guess you need to find a place for this function. Ultimately it should be in logging.h logging.cpp because logging is the only user of this function. But right now the logging PR #309 is unfinished, to be delayed after this PR. To avoid conflict, I think it's ok to temporarily put static double getTime() and the whole definition in libfreenect2/packet_processor.h. This creates a bit duplicate code, but the logging PR will promptly move it back into logging.cpp.

[fran6co]

Made a new Timer class, when #309 is merged it can be moved to the logging source files.

[xlz]

The rest looks fine to me.

[fran6co]

Now it's using the full resolution of chrono.

[xlz]

Tests OK on Intel/Linux/OpenCL.

[floe]

Alright, last big one. I'm pretty certain it will still break in some contexts, but if we don't merge it now, this release will never happen :-)