DEP: Remove normed= keyword argument from histograms

[seberg]

The normed keyword argument has been deprecated for a long time.
This removes it, replacing its position with the new density
argument.

[seberg]

I decided to replace the positional argument. In theory, that changes behavior in those cases where normed was buggy before (it thus should have had a FutureWarning rather than a DeprecationWarning).

If anyone has any concerns about that, we could potentially escalate or make the release note a bit clearer on that front.

[rossbar]

If anyone has any concerns about that, we could potentially escalate or make the release note a bit clearer on that front.

I think the only case where the behavior would change would be if a user called the function with positional args and non-uniform bins, in which case the previous behavior was buggy and the new behavior is correct, so this seems fine to me 👍

[charris]

Thanks Sebastian.

[tacaswell]

The 1D case was well documented + warnning, but the 2D and nd cases appear to only have been documented as deprecated by reference and never warned.

[seberg]

Hmmm, true, not sure I really realized this when doing this. Do you think we should "go back" to a deprecation for those?

[tacaswell]

I got back-channel grumpy users claiming "no one in Python cares about back-compatibility" over this so maybe.

However, given we are now over a year and 2 minor versions past this I'm not sure how many people who will be bit by this are left and of those how many won't be bit by it anyway (if they are lagging versions they likely won't jump to the newest version but go to 1.24 eventually).

[seberg]

Oh, missed that there are 2 minor versions out there, so even the normal deprecation wouldn't have made a difference... I guess it is pretty much too late anyway, then.