Adding unit checking/conversion to entry points

[chrisiacovella]

As discussed in our roadmap, we want to have units associated with entry points.

Two key spots in the code that require this:

• loading hdf5 files

  • The use of the curated datasets should already ensure proper units, but checking the units against expected ones would provide an additional consistency check.

• The expected units should be defined in the context of the 'available_properties' for a given dataset (i.e. convert the list of properties to a dict).

  • If checking units is prohibitively slow for all records, we can have the default behavior be to simply check the first entry (full checking can be enabled by an optional flag)
  • We should print to the log and screen the specific units that are being used for clarity.
  • Since we are using curated datasets, I do not think we need to implement unit conversions (although this is would be trivial); this is mostly to ensure we do not have inconsistencies

• passing cutoffs (e.g., distance cutoffs, force cutoffs):

  • Rather than just encoding the expected unit in the doctoring, we should require a user to assign the units to the variable as an openff-unit quantity (e.g., 0.5*unit.nanometer).
  • In this case I think we should absolutely support conversion to the appropriate unit (with a warning/note to the user that we are converting to the expected unit)

[wiederm]

Part of this has been addressed in PR #56

[wiederm]

In my opinion, this has been addressed in the linked PRs. Closing for now.