GNIRS HR-IFU "sacrificial flat" quirk

[phirstgemini]

Per Email from Ema 2024-04-15:

GNIRS flat association needs a tweak to handle a GNIRS quirk.

Basically they will take junk files that are labeled as acqCal FLAT with HR-IFU. These files should not be associated as calibrations.

There may be implications on how we select the actual flats here too... ie we probably can't associate flats over prism position tweak boundaries, but we likely have no idea where those are...

---

Dear all,

We are updating the OT templates for the upcoming event of the HR-IFU! 
Given the non-reproducibility (at the level of accuracy that is needed) of the GNIRS prism position, before starting an acquisition we will need to take one/two "sacrificial flats" (~18 seconds each) to make sure that the HR-IFU footprint is fully into the detector. 

The procedure will be:
1. start slewing and start the process of getting the sacrificial flat in place
2. take the sacrificial flat and measure the x-shift on the detector with AndyS' magical script
IF the x-shift is relevant (greater than ~15 pixels):
    3. update the LUT for the prism using AndyS' even more magical script
    4. take a sacrificial flat to confirm that the footprint is as expected.
    5. go to the acquisition sequence
ELSE
    3. go to the acquisition sequence

If the class is set to Automatic, these will be "Nighttime partner calibrations", however, these are not "science" flats and they are going to be taken in a sequence that is different from the "Obs: Science" one (and likely in a different configuration to allow for the short exposures). 
The proposal is to switch the class to "Acquisition Calibration", to make it clear that these are not related to science (and should not be used for data reduction), but are just a step necessary to properly acquire and take the science exposures.

Please let me know if you see any issue with this (or if you can propose any better way to differentiate these flats).

Cheers and all the best,
Ema

[phirstgemini]

This has been implemented, but there's no actual test data yet.

[phirstgemini]

Asked Ema via slack on 9-sep-2024 if any example data exist...

[phirstgemini]

One set of example data exists: GN-2024B-Q-322-46-001 = N20240906S0043.fits is the sacrificial flat

[phirstgemini]

Except it turns out there's no corresponding science data, so it's not possible to test fully that the flat doesn't get associated. Pushing this to 3.3...