Telescope area always uses the largest mirror

[teutoburg]

When "!TEL.area" isn't set in the cmds (as is the case e.g. in MICADO calib mode), the OpticsManager calculates it from the largest mirror area. This is usually correct, e.g. when that largest mirror is the ELT M1. But in some cases (e.g. MICADO calib mode), there is a slightly larger mirror after a smaller one, so the first one should be used instead. If all mirrors are in the correct order, we should be able to default to the first one instead of the largest one?

[teutoburg]

In some way related to #692

[oczoske]

The automatic determination of "telescope area" is very dangerous. For instance, for MICADO calibration neither the largest nor the first optical element give the correct area - that is actually the ELT M1 area! The decisive quantity is the etendue, the product of area and solid angle, which is conserved through the optical system. Opening angles of the beam and beam cross-sections change, but the product remains the same. We compute etendue at some point where it is convenient to do so, e.g. at M1 by multiplying the M1 area with the solid angle that a pixel sees on sky (etendue per pixel, if you will). Since the etendue is the same at the detector (pixel area times solid angle under which it sees the last optical element) regardless of whether we use the telescope or the MCA, hence the etendue is the same for the calibration mode as it is for on-sky mode. Since we give "position" in calibration mode in arcsec as if we were looking on-sky, we need M1 area to get the etendue.
Long story short, !TEL.area should be set in the configuration yaml for non-telescope modes, as it is done for METIS and now (hidden in AstarVienna/irdb#372) also for MICADO.