Rewrite sample_data.py as a class

[niksirbi]

Is your feature request related to a problem? Please describe.
Currently the functionalities for fetching sample datasets are implemented in the sample_data.py module. This works well but the interface results in some awkwardly worded and repetitive expressions, e.g.:

from movement import sample_data

sample_data.list_sample_data()
file_path = sample_data.fetch_sample_data_path("filename")
ds = sample_data.fetch_sample_data("filename")

Describe the solution you'd like
Perhaps it's better to refactor the functionality into a class named SampleData and have the interface look sth like below:

from movement import SampleData

sample_data = SampleData(local_data_dir)
sample_data.list()
file_path = sample_data.fetch_path("filename")
ds = sample_data.fetch("filename")

The current global variable DATA_DIR could be initialised as class attributes local_data_dir with the default value ~/.movement/data, but giving the user the option to send it elsewhere.
The global DATA_URL could be made into a class variable with a fixed value.

Context
This is not critical or urgent, just wondering which interface feels more intuitive and consistent to people.

[niksirbi]

Look into how scikit-learn, scikit-image, napari, and nilearn do it.

[niksirbi]

Consider SampleDataRegistry as a name for the class.