Prism 2d correct center calculation

[Yaraslaut]

Fixes #2144

[codecov-commenter]

Codecov Report

Merging #2188 (0b7fad6) into master (2aa9164) will decrease coverage by 8.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #2188      +/-   ##
==========================================
- Coverage   73.09%   65.07%   -8.02%     
==========================================
  Files          17       17              
  Lines        4906     4908       +2     
==========================================
- Hits         3586     3194     -392     
- Misses       1320     1714     +394     

Impacted Files	Coverage Δ
python/geom.py	94.16% <100.00%> (+0.02%)	⬆️
python/adjoint/wrapper.py	24.61% <0.00%> (-73.85%)	⬇️
python/adjoint/filter_source.py	23.07% <0.00%> (-65.39%)	⬇️
python/adjoint/objective.py	41.50% <0.00%> (-51.50%)	⬇️
python/adjoint/utils.py	48.83% <0.00%> (-32.56%)	⬇️
python/source.py	64.06% <0.00%> (-31.25%)	⬇️
python/simulation.py	71.98% <0.00%> (-4.88%)	⬇️
python/adjoint/optimization_problem.py	53.68% <0.00%> (-3.16%)	⬇️
python/verbosity_mgr.py	65.15% <0.00%> (-1.52%)	⬇️
python/adjoint/filters.py	60.14% <0.00%> (-0.73%)	⬇️

[smartalecH]

I retriggered the tests now that the MPB bug was fixed, but it's still failing (ignore the fact that the first batch of tests is passing... that always passes so long as it builds correctly and we need to fix the test matrix...)

The output of test_prism.py might help you debug:

Using MPI version 3.1, 2 processes
Testing Non-Convex Prism #3 using marching squares algorithm...
-----------
Initializing structure...
Splitting into 2 chunks by voxels
non-coplanar vertices in init_prism
non-coplanar vertices in init_prism

(here's a related issue from libctl)