[Feature request] Create an object that loads NWB datasets as a `DataFrame` or `DataArray`

This object should search for any datasets within the NWB file and make them accessible via dot notation (substituting underscores for any spaces in the name). For example, if an NWB file contains multiple datasets within `acquisition` or `processing` groups, these can be loaded by typing e.g., `session.Probe-A-LFP`, instead of using the full path within the file. We just have to be careful about potential naming conflicts, in case multiple datasets have the same base name, but different paths.