Hisanori Furusawa has identified additional HSC defects that should be masked. He writes:
In this updates, I have added a mask to ccd=76 as well as ccd=33 and
43, as we found some affected area by bad pixels especially in NB0387.
Also I removed an unnecessary mask in ccd=43 which was mistakenly
placed for early period (before 2016-04).
The top-left corner of ccd=33 and 43 look quite bright on dark images
in some periods, but we determiend the areas to be masked on CORR
images both for typical broad-band (mostly HSC-G), and the worst