Will concurrently reading chunks within a single variable read speed up access

[jpswinski]

Currently, the code is only concurrent at the variable level, meaning that different variables read from an hdf5 file are read concurrently.

For variables that contain a lot of data (e.g. ICESat-2 ATL03 heights group variables), it may be beneficial to break up the reading of the chunk data into concurrent threads. Even better would be the ability to process those chunks in threads that executed on different cores.