Remove Observable_stat external linkage

[jngrad]

Description of changes:

• re-use code in the cython interface to extract data from Observable_stat objects
• remove obs_scalar_pressure global variable
  • write-only variable in the core -> can be calculated from obs_pressure_tensor in cython
  • simplify pressure kernels
  • less MPI communication
• rename obs_pressure_tensor to obs_pressure for consistency with obs_energy
• remove external linkage for obs_pressure and obs_energy
  • these globals are no longer visible outside of their translation unit
  • use a const reference getter for the script interface
  • partial fix for Remove global variables #2628
• narrow includes of Observable_stat.hpp
• add ELC/DLC energy corrections to the P3M/DP3M/DDS solver energy slots
  • the analyze.energy(), analyze.pressure(), analyze.pressure_tensor() functions now only reports 2 slots for Coulomb and dipolar contributions: particles contribution and electrostatics/magnetostatics solver contribution

[codecov]

Codecov Report

Merging #3770 into python will increase coverage by 0%.
The diff coverage is 96%.

@@          Coverage Diff           @@
##           python   #3770   +/-   ##
======================================
  Coverage      89%     89%           
======================================
  Files         551     551           
  Lines       24468   24462    -6     
======================================
- Hits        21812   21807    -5     
+ Misses       2656    2655    -1     

Impacted Files	Coverage Δ
src/core/constraints/ShapeBasedConstraint.cpp	88% <ø> (ø)
src/core/electrostatics_magnetostatics/dipole.cpp	65% <72%> (+1%)	⬆️
src/core/Observable_stat.cpp	100% <100%> (ø)
src/core/Observable_stat.hpp	100% <100%> (ø)
src/core/electrostatics_magnetostatics/coulomb.cpp	78% <100%> (ø)
src/core/energy.cpp	97% <100%> (+<1%)	⬆️
src/core/energy_inline.hpp	84% <100%> (+<1%)	⬆️
src/core/pressure.cpp	97% <100%> (-1%)	⬇️
src/core/pressure_inline.hpp	81% <100%> (-2%)	⬇️
src/core/particle_data.cpp	96% <0%> (-1%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 030a97a...88b63bd. Read the comment docs.

[fweik]

Clear improvement. I think some of the output parameters should be avoided in the cases where that is simple, see inline comments. This helps to reduce coupling.

[fweik]

I think you can remove some more includes so that we also have physical decoupling

[fweik]

LGTM