closes #13244

There have been several iterations on the issue of eyetracking units, so let me summarize for brevity:

Stemming from Issue #12756 (comment) (and PR #12846 (comment) + #12850), we changed MNE internals to assume all incoming Eyetracking data were already represented in Meters (SI unit of distance, as some eyetrackers report pupil size in mm). In those PR's I adjusted the scaling for Pupil channels so that data in meters would look nice when plotting by default. I chose a scaling factor of 1e6 (meters -> micro meters).

This caused inst.to_data_frame() to convert from meters -> micrometers which is probably not very intuitive for researchers (hence #13244).

It is probably more intuitive to adjust the scalings to go from meters to mm. The plotting still looks good, and pupil channels will be converted to mm by default by to_data_frame():

import mne
import numpy as np


sfreq = 100
ch_names = ["my_pupil"]
ch_types = "pupil"
info = mne.create_info(ch_names, sfreq, ch_types)
data = np.full((1, 900), np.repeat(np.linspace(0, .008, 9), sfreq))  # data between 0-8mm
raw = mne.io.RawArray(data, info, verbose=False)
raw.plot(remove_dc=False)

df = raw.to_data_frame()
df.head()

Out[4]: 
     time  my_pupil
0    0.00       0.0
1    0.01       0.0
2    0.02       0.0
3    0.03       0.0
4    0.04       0.0
..    ...       ...
895  8.95       8.0
896  8.96       8.0
897  8.97       8.0
898  8.98       8.0
899  8.99       8.0

[900 rows x 2 columns]

Does this sound reasonable to you @larsoner and @drammock ?