Update OverlapCalculator.py

[ptrab]

adding the functionality to use the Kuhn-Munkres algorithm to update the tracked root.
can be activated through the calc-argument min_cost: True and default is False currently.

---

As an example, let's assume we are tracking excited state number 8 (old 8) and in row 7 and 8 and also column 7 and 8 of an overlap matrix between two optimization steps, there are the following values:

       new 7 | new 8
old 7 | 0.65 | 0.71
old 8 | 0.54 | 0.56

For old 8 we see that new 7 and new 8 are very similar and, based on the current algorithm, we would choose new 8 because of its higher overlap (56% vs. 54%) in this row and, therefore, not update the root. However, if we take a look at the column for new 8, we see that old 7 and new 8 have a much higher overlap of 71% (vs. 56%). As such, new 8 is more likely to be old 7 than old 8 and vice versa.

Using the Kuhn-Munkres algorithm, each old excited state gets mapped to only one new excited state based on the whole overlap matrix which should be more reasonable than taking only a single row into consideration while ignoring all other information.

---

I did not outsource the new code into a new function ... sorry :D

[eljost]

Please capitalize the beginnings of the comments.

[eljost]

add one space after reasonable

[eljost]

Delete this line and move (the same) line 803 outside the else block.

[ptrab]

Currently it is also highly confusing in these few lines of code as col_ind and row_ind are very similar to col_inds and row_inds but should not be confused. The "singular" forms are the ones from the current code, while the "plural" forms hold the results from the new linear_sum_assignment.