[BUG] `mne.label.find_pos_in_annot` doesn't check `coord_frame`

It doesn’t seem like a good idea to allow any position to be passed to find_pos_in_annot without checking that it is in the mri coordinate frame. If someone has the wrong coordinate frame, the error will be silent and they will get the wrong ROI which to me seems like a pretty good risk for incorrect results that could be avoided by enforcing the coordinate frame or maybe rather making this a class function of Dipole.

[larsoner]

I don’t think Dipole would be the right subclass, in general the point you’re talking about could be anything (iEEG sensor, part of a mesh, an elementary current dipole, etc.). I think we should do two things:

1. Document clearly that if an ndarray or list-like is passed, these must be in the MRI coordinate frame
2. Add support for other objects as we need them, such as DigPoint instances (?), where we can check the coordinate frame

FWIW Dipole always lives in the head coordinate frame, so if we wanted to add support for that in (2) then we’d have to have a trans argument in that function, if we don’t already.

But really I think we should do the doc update (1) first and then let actual use cases guide (2)

[larsoner]

Dipole is always in head coords and I don’t think it’s a common use case anyway. Let’s just start from montage (and not worry about Dipole) since it’s the use case we actually want to support for iEEG?