Fixing multiplane

[AlexandreAdam]

coauthor: @andreasfilipp

We are fixing the math of the multiplane class based on Fleury et al. (2022). Doing this, we realize that lenstronomy must be wrong. The multiplane equation is actually much simpler than we thought. It's just recursively applying the single thin lens equation, while updating the coordinate.

All the normalization quantities, like $\Sigma_{\mathrm{crit}}$, are computed by taking the next plane as the source plane. This is how the math stays internally consistent.

Also fixed the Shapiro time delay units, from days * arcsec^2 to days. A conversion factor was missing.

This PR also attempts to fix the circular logic in the ThinPlane class by forcing inherited class to implement the reduced_deflection_angle method. The physical_deflection_angle method is then just a wrapper on the previous one. See Issue #242.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 93.14%. Comparing base (9806a84) to head (79888d3).
Report is 41 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #244      +/-   ##
==========================================
- Coverage   93.22%   93.14%   -0.09%     
==========================================
  Files          59       59              
  Lines        2732     2741       +9     
==========================================
+ Hits         2547     2553       +6     
- Misses        185      188       +3     

Flag	Coverage Δ
unittests	93.14% <100.00%> (-0.09%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.