New benchmark for data object creation

[Jonathanlyj]

Add a benchmark program that creates a large number of data objects and writes significant metadata to the output file. It distributes subsets of data objects to processes, and it uses MPI communications so that all processes share consistent metadata for variables, dimensions, and attributes.

[wkliao]

To see the errors from failed GitHub workflow runs,
please run the following commands on you local machines.

make check
make ptests
make distcheck -j 8 CFLAGS="-Wno-alloc-size-larger-than"

See DEVELOPER_NOTES.md for more information about testing.

[wkliao]

In file benchmarks/C/indep_data_obj_create.c, please add a comment block
at the top to give a detailed about the new benchmark.
Check examples in aggregation.c, and pnetcdf_put_vara.c

[wkliao]

If there is a timing output on screen, it will be nice to add it to the comment block.