Be more explicit about what is returned by read_epochs_eeglab and read_epochs_kit

[nabilalibou]

Proposed documentation enhancement

It's a minor detail, but the docstrings for functions like mne.read_epochs_eeglab() and mne.read_epochs_kit() indicate that an Epochs instance is returned, while in reality they return subclasses of BaseEpochs; specifically EpochsEEGLAB and EpochsKIT, respectively.
It can be deceptive, for example it could encourage the use of a isinstance(epochs, mne.Epoch) because although BaseEpoch exists, the documentation could misleads into thinking that there is only one type of Epoch class.

Even though there don't seem to be any html pages created for these subclasses that could be linked, i think we could add a few details to be clearer, what do you think ?

[hoechenberger]

this problem affects many readers and I believe it has at least been brought up when we added some type hints not too long ago.

i agree with you that this should be improved.

parts of the class hierarchy would probably need to be revamped.

[larsoner]

For now the easiest solution would be to say that the return type is BaseEpochs, which is nice because it is a documented type

[nabilalibou]

A solution in the meantime to make sure the reader has all the keys needed to understand what's behind could be something like:

Returns:
   EpochsEEGLAB/EpochsKIT : instance of [BaseEpochs]
        EpochsEEGLAB/EpochsKIT object (subclass of [BaseEpochs])

And having a markdown box:

See also: 
   [mne.Epochs] Documentation of attributes and methods.