Add a function to copy data between points of different types

[taketwo]

This pull request adds a new function that allows to copy data between points of different types:

template <typename PointInT, typename PointOutT> void
copyPoint (const PointInT& point_in, PointOutT& point_out);

The function has several specializations, which the compiler instantiates depending on the particular combination of input and output point types.

There are three cases:

• Points have the same type.

  In this case a single memcpy is used.

• Points have different types and one of the following is true:

  • both have RGB fields;
  • both have RGBA fields;
  • one or both have no RGB/RGBA fields.

  There is no RGB/RGBA mismatch involved, so we simply find the list of common fields and copy their contents one by one with NdConcatenateFunctor.

• Points have different types and one of these types has RGB field, and the other has RGBA field.

  In this case we also find the list of common fields and copy their contents. In order to account for the fact that RGB and RGBA do not match we have an additional memcpy to copy the contents of one into another.

All decisions are made at compilation time, so at run time only the copy instructions are called.

This pull request also adds a unit test for copyPoint() and updates the family of copyPointCloud() functions to use it. This dramatically simplifies them.

Finally, I searched through the PCL code to identify places where the authors intended to copy data between different point types, and replaced it with copyPoint() calls. I should note that in all cases the existing code did not account for RGB/RGBA mismatch, so this update potentially fixes some not-yet-discovered bugs for certain pairs of point types.

[rbrusu]

That is an AWESOME contribution Sergey! Give me a few hours to mull it over, but you won't see any resistance from my behalf :) 👍

[taketwo]

By the way, we can also instantiate the template for the catesian product of all point types. Do you think it makes any sense?

[sdmiller]

This is fantastic! It simplifies so much code, and provides functionality I've found myself manually re-writing far too many times. Really solid contribution!

The logic in the code seems sound to me. As for the instantiation question, I can't think of a good reason not to for such a small (but fundamental) building block. Radu?

[jpapon]

I always wondered why this function didn't exist - and why I was forced to do workarounds using copyPointCloud.
Great job! This is a really useful addition!