gegl: use a 1e-11 instead of 1e-6 for matrix coefficient epsilon

This should help with some problems observed in issue #261.
parent 8fbdf0c5
Pipeline #200531 passed with stages
in 6 minutes and 38 seconds