Better handling of montage files

[jasmainak]

When reading montage files, users seem to face a number of issues.

1. Documentation of the different file formats is not clear. For example, *.hpts should not have a header and a column of type. This is not clearly explained anywhere. You cannot figure this out without looking at the code.

2. Inconsistent handling of numpy versions when reading text files. Here it is handled properly: https://github.com/mne-tools/mne-python/blob/master/mne/channels/montage.py#L325-L328 but this is not the case in all calls. Ideally, this should be abstracted away into a private helper function.

maybe an easy issue for @fmamashli to solve who encountered this in the first place

[agramfort]

FYI the read_montage is currently being deprecated in #6764 to make sure we are addressing the issue @fmamashli <https://github.com/fmamashli> can you clarify the usecase? How was the .hpts file produced? what does it contain? points in what coordinate system?

…

[massich]

we bumped the minimum version of numpy. We no longer need to keep both versions of skip_row / header.

[fmamashli]

We digitized the channels using a new system: Localite.
.hpts: we followed MNE manual: EEG, channels name, coordinates (x,y,z) in mm.
The coordinate system is RAS.

The problem was that in MNE-python, there was no good documentation on the format of the files and what exactly it should contain. like headers, columns, coordinate system.

[jasmainak]

also just to add to what @fmamashli said, see the documentation of mne-c. It is clearly mentioned what the format should look like. Otherwise, you start seeing obscure errors from np.loadtxt which are hard to debug

[agramfort]

if you use the current master you can try this command: https://mne.tools/dev/generated/mne.channels.make_dig_montage.html#mne.channels.make_dig_montage provided you have the locations already read in python you could ideally specify that the points are in MRI RAS and then it should work just fine (hopefully). share some fake data if needed.

…

[jasmainak]

We managed for now. But I would say you need a documentation page for montages analogous to this one for reading in raw files.

[agramfort]

should be fine now with: https://mne.tools/dev/auto_tutorials/misc/plot_sensor_locations.html#sphx-glr-auto-tutorials-misc-plot-sensor-locations-py https://mne.tools/dev/overview/implementation.html#dig-formats

[jasmainak]

Cool !