I3454 missing mpi functions in doc

[dmitry-ganyushin]

Corrected Doxyfile.

[eisenhauer]

Target for release_29?

[eisenhauer]

There may be some confusion between ADIOS2_HAVE_X and ADIOS2_USE_X because there's some differences in meaning internally and externally. In building ADIOS, ADIOS2_USE_X is used to control whether or not we should use X in the build and mostly controls CMake. I.E. if ADIOS2_USE_X is set to ON, CMake looks for X and it's an error if it doesn't find it. If set to Auto it looks for X and it's not an error if it doesn't find it. If set to Off, it doesn't look for X. Then ADIOS2_HAVE_X is set to true if USE was ON or AUTO AND the thing was found. So in the core code, I think we should be generally looking at HAVE_X, not USE_X. However, the bindings are a bit different. There we expect the user to define ADIOS2_USE_X if they want to build with/use a version of ADIOS that was built with X. So, we can have conditionals in the bindings that look at, for example ADIOS2_USE_MPI, where we probably don't want to do that in core code. I think this was different originally and perhaps Chuck formalized that distinction. That change, and spots where it was missed, is probably the source of this problem, and it would probably be wise for us to maybe do a quick audit to see if there are other places where this isn't properly done.