Fix division error in inactive leaf size estimation

[kulbachcedric]

Hi all,
just added the check for devision errors as mentioned in #1717

Best
Cedric

[e10e3]

Hello Cedric,

Are you sure this is the correct fix?
The issue mentioned a problem with the number of active leaves (the line above).

[kulbachcedric]

Hi @e10e3
you're right, but are we not eventually facing the same issue in line 301?

[smastelini]

LGTM! Thanks @kulbachcedric to look into that! Would you mind adding an entry in the release notes?

[smastelini]

Sorry! I didn't see the use of the active leaves in the lines above.

[e10e3]

are we not eventually facing the same issue in line 301?

Yes, we are, you're correct.
The code is similar, and the same problem could happen with the next division, so both line 299 and line 301 can benefit from the same fix you propose.

Your first commit was not wrong per se, but because you mentioned that you just applied the suggestions from the issue reporter, I wanted to point out that the issue was about a different line, so that you didn't end up patching a different problem while keeping the original one unsolved 😁.

In the end, both lines are affected by the same bug and a complete fix would apply to both.

[e10e3]

Thank you for adding a changelog note. Don't you think it should go in another section? Previous changelogs had a section for each module, so it would make sense to sort this change under “tree” rather than “stats”.