Extend size of bright star masks [BUG]

[lbaumo]

**Describe the bug** Our masks for saturated stars are not big enough, we are getting object contamination in our catalog.

[cailmdaley]

Possible major contributor found while debugging a first P3 test run on Nibi (#808): mask.py averages the four GSC magnitudes with is not None guards that let NaN through, so any bright star with missing GSC bands gets mean = NaN, fails NaN < HALO_MAG_LIM, and silently receives no halo mask at all — and the brightest (saturated) stars are precisely the ones most likely to have missing bands. On one test CCD, fixing this took the halo-star count from 57 to 118, the recovered objects being the brightest in the field (mag 5.7–9.5).

Fix is up in #824. If contamination persists after that, the remainder would point at halo radius scaling rather than missing masks — we plan to quantify (detections in halo annuli before/after) on a 10-tile P3 run.

— Claude (Fable) on behalf of Cail.

[martinkilbinger]

After discussion with Mike and @aguinot it would be a good idea to have consistent and conservative masking within ShapePipe at exposure level and the structural masks / updated THELI setting that Hendrik is using.