namedtuple is incompatible with num_workers>0 in PyTorch DataLoader

[KonstantinWilleke]

The __getitem__ of the datasets can not handle a namedtuple when there are multiple parallel workers. The parallel workers are needed to reach high data loading speeds for powerful compute nodes.
This is a fundamental pytorch issue: because each worker instantiates a Dataset object, and the namedtuple is instantiated in each one, the parallel workers can't collate the batches in a "custom" named tuple.

Potential workarounds:

• write a custom collate_fn for the DataLoader which converts the entire batch into a namedtuple.
• each batch returns a dictionary, as in {device_name: value_tensor}

[pollytur]

Another potential solution is to do same as in neuralpredictors and define the namedtuple globally, then it works with the num_workers>0

[pollytur]

this is still valid
named tuples needs to be removed from

• experanto/experanto/experiment.py

  Line 5 in ee767fb

  from collections import namedtuple

• experanto/experanto/datasets.py

  Line 7 in ee767fb

  from collections import namedtuple

• self.DataPoint also needs to be removed -

  experanto/experanto/datasets.py

  Line 55 in ee767fb

  self.DataPoint = namedtuple("DataPoint", self.device_names)

• I am not sure about TimeInterval -

  experanto/experanto/intervals.py

  Line 7 in ee767fb

  class TimeInterval(typing.NamedTuple):

  - we do not use it so far in the data loading but might become an issue later

[pollytur]

ok, everything apart from class TimeInterval(typing.NamedTuple): was addressed in #90 so I will close it here