Dataset name and root directory when loading from existing dataset

Issues

Issue 1 - solved with !42 (merged)

Root directory seems to be saved in the dataset file as a local path. When the whole package is moved to a different location, dataset cannot be restored because it looks for files in a wrong (non-existent) location.

Issue 2

Currently, the dataset name and the folder name in which the data sample and the data object reside and the dataobject file name are strictly linked and inscribed in the pickled dataobject.

Required

The user should be able to move the whole package around, pass it on to other users/machines, and rename the top folder ad libitum.

Clarify:

• is there a scenario where we have different dataset objects for the same "dataset" (collection of data samples)?

Proposed solution

Issue 1

In Dataset class:

• in self.save_dataset_obj: reset self.datapath = None
• in self.load_dataset_obj: update the self.datapath with the current path

Issue 2

• folder name and dataset name should be separated. Locate a pickled/json'd dataset object using an explicit path.

Further proposition:

Probably a more elegant solution would be if the user could load the Dataset object from explicit file path.

Now (user needs to know the correct name of the dataset):

dataset = Dataset(name='myname', path=<mypath>)
dataset.load_dataset_obj()

Proposed:

dataset.load_dataset_obj(path=<mypath>)

(the dataset name is saved in the object)