use_fast_tensor_data_set breaks during_training_metric

[elcorto]

When adding this

parameters.running.during_training_metric = "band_energy"
parameters.data.use_fast_tensor_data_set = True

to examples/basic/ex01_train_network.py, we get

Traceback (most recent call last):
  File "/path/to/mala/examples/basic/ex01_train_network.py", line 95, in <module>
    test_trainer.train_network()
  File "/path/to/mala/mala/network/trainer.py", line 424, in train_network
    vloss = self.__validate_network(
            ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/path/to/mala/mala/network/trainer.py", line 1149, in __validate_network
    self._forward_entire_snapshot(
  File "/path/to/mala/mala/network/runner.py", line 399, in _forward_entire_snapshot
    (data_set[from_index:to_index])[1], as_numpy=True
     ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^
  File "/path/to/mala/mala/datahandling/fast_tensor_dataset.py", line 40, in __getitem__
    idx * self.batch_size : (idx + 1) * self.batch_size
    ~~~~^~~~~~~~~~~~~~~~~
TypeError: unsupported operand type(s) for *: 'slice' and 'int'

[RandomDefaultUser]

Hi @elcorto I just tested this and while I can reproduce this issue with an older version of this code I no longer see this error with commits starting from #560, which severely overhauled the reporting/validation process. I would close this issue for now, but if this issue arises again or the fast tensor data set behaves differently than expected after these changes feel free to create a new one.