adjust arc spectro error thresholds to avoid false positives

[sbailey]

Observers have been reporting Arc spectro errors every night, from the QA Status "spectro" column of the exposure list per night, e.g. https://nightwatch.desi.lbl.gov/20240630/exposures.html . These errors appear to be coming from individual camera line measurements that are slightly out of normal.

I think the primary purpose of these checks is to see if a lamp is burned out or if there is major contamination. As such, the warning/error thresholds could be significantly expanded to avoid false positives.

[sybenzvi]

Following the discussion we had on slack with @jose-bermejo, I went ahead and updated the line references and widened the warning and error thresholds. Tests on two exposures from 20240706 look fine: https://data.desi.lbl.gov/desi/users/sybenzvi/nightwatch/20240706/exposures.html

I'll do a bit more cleanup and then merge this change.