Back-calculate exposure times for BACKUP tiles

[geordie666]

We installed a new version of the ETC on the mountain (see the discussion in desihub/desispec#2370). We now need to update previously observed BACKUP tiles to align with the new ETC calculation (see, e.g., desihub/desispec#2370 (comment)).

This may need to be done on the desispec side, in which case we can open a second issue there. But, for now, we'll track this in surveyops.

[sybenzvi]

During a discussion with the data team in today's survey-ops telecon, it's not clear if a simple tsnr -> EFFTIME update is possible for the BACKUP tiles. We might try to reprocess all BACKUP tiles from scratch, but this is a nontrivial and time-consuming exercise, and assumes the oldest BACKUP tiles can be easily reprocessed in the first place.

@akremin will check one of the older tiles to see if the tsnr tables can be updated, or if reprocessing is needed and possible.

Depending on the outcome of this check, we might need help from users of the BACKUP program to recalculate EFFTIME for the BACKUP tiles.

[schlafly]

FWIW, here's the list of tiles and exposures we should update:
https://data.desi.lbl.gov/desi/users/schlafly/misc/tiles-backup-needing-efftime-updated.ecsv
https://data.desi.lbl.gov/desi/users/schlafly/misc/exps-backup-needing-efftime-updated.ecsv
258 exposures on 105 tiles, spanning 81 nights.

Depending on what is easiest, I'd be happy e.g. if we wanted to skip updating nightqa and instead only reprocessed the tiles.

It seems like a horrible hack and doomed to cause problems and confusion, but Stephen has requested a more "surgical" update a few times. One hacky option there would be to just multiply the effective times of those tiles in tiles-daily by 10^(2 * 2.165 * EBV / 2.5). That won't be strictly correct but it will be better than what's in there now.