PaK: solution for singular likelihood hessian and infinite volume shells

[diegodoimo]

I propose to solve some pathologies that actually result in 'None' values in the density estimated with pak:

1. presence of shells of infinite volume: here I assign a volume equal to e^300 which is almost the largest floating point number. I didn't want to set it to the largest value possible because the volumes are used in the max likelihood routine, and I didn't test the behavior of using np.inf there;

2. singularity of the hessian matrix in the optimization of the likelihood: here I added a fixed point step whenever the hessian is singular. This 'bug' is particularly nasty as it leads to none values without specific error messages.

[codecov-commenter]

Codecov Report

Merging #77 (8296d3f) into main (f2af10a) will decrease coverage by 0.16%.
The diff coverage is 68.75%.

@@            Coverage Diff             @@
##             main      #77      +/-   ##
==========================================
- Coverage   78.16%   78.00%   -0.17%     
==========================================
  Files          10       10              
  Lines        1177     1191      +14     
==========================================
+ Hits          920      929       +9     
- Misses        257      262       +5     

Impacted Files	Coverage Δ
dadapy/_utils/density_estimation.py	56.57% <68.75%> (+1.74%)	⬆️

---

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 f2af10a...8296d3f. Read the comment docs.