Use grid_stride_range in kernel loops

[harrism]

Closes #650

Description

Adds a dependency on Ranger and uses ranger::grid_stride_loop with a range-based for loop in outer loops in every raw kernel in cuSpatial. In the few cases where we had raw kernels without grid-stride loops, this adds one. Also adds guidance to the C++ developer guide.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[isVoid]

Thanks!

[vyasr]

Do we want to move ranger into the rapidsai org or somewhere else shared before we introduce a dependency on it here?

[harrism]

Do we want to move ranger into the rapidsai org or somewhere else shared before we introduce a dependency on it here?

I am not opposed to that. That would make it easier for e.g. cuDF to adopt it.

[harrism]

Decision is to merge this as-is for now. We can look at re-homing ranger if and when it is determined to be needed.

[harrism]

/merge