Investigating gain selection issue

[vuillaut]

I open the thread to follow the gain selection issue raised by PR #72.

It seems I could reproduce an (the?) issue in the case of waveforms of shape (n_gain, n_pix, 1) - that would correspond to ASTRICam.
In this case the gain_mask from _, gain_mask = gain_selector.select_gains(cam_id, waveform) is of shape (n_pix,).

Maybe @riwim or @mackaiver can tell me if that's the issue they encountered.

[TjarkMiener]

Solved by the ctapipe process tool.