Fix RuntimeWarning in singlediode when v_oc is zero

[Blizzeq]

• Closes RuntimeWarning: invalid value encountered in divide when using singlediode() #2671
• I am familiar with the contributing guidelines
• I attest that all AI-generated material has been vetted for accuracy and is in compliance with the pvlib license
• Tests added
• Adds description and name entries in the appropriate "what's new" file in docs/sphinx/source/whatsnew for all changes. Includes link to the GitHub Issue with :issue:`num` or this Pull Request with :pull:`num`. Includes contributor name and/or GitHub username (link with :ghuser:`user`).
• Pull request is nearly complete and ready for detailed review.
• Maintainer: Appropriate GitHub Labels (including remote-data) and Milestone are assigned to the Pull Request and linked Issue.

With v_oc = 0 the bracket passed to find_minimum was (-1, 0, 0), which breaks the x1 < x2 < x3 requirement @jwhitaker-gridcog pointed at, and scipy warns from a 0/0 inside the solver. Instead of silencing the warning, the right end of the bracket is 1 wherever v_oc is 0. Current falls with voltage and is zero at v_oc, so -v*I(v) is positive on both sides of 0 and any positive right end still brackets the minimum there.

Results are unchanged wherever v_oc > 0. At v_oc = 0 you still get v_mp of about 0 and p_mp = 0, now without the warning. The new test turns warnings into errors and fails on main; it skips on scipy older than 1.15, which takes the golden section path.

[github-actions]

Hey @Blizzeq! 🎉

Thanks for opening your first pull request! We appreciate your
contribution. Please ensure you have reviewed and understood the
contributing guidelines.

If AI is used for any portion of this PR, you must vet the content
for technical accuracy.

[cwhanse]

I don't agree with enlarging the interval by assigning a positive value to the right endpoint. That risks returning a small vmp>0 which doesn't respect the known inequality vmp <= voc.

At v_oc = 0 you still get v_mp of about 0 and p_mp = 0

[Blizzeq]

Agreed, thanks. You said the same in the issue and I missed it. I measured it: with the positive right end, v_mp came back as 9.9e-17 at v_oc = 0. The bracket (-1., 0.8*v_oc - 0.2, v_oc) from the issue isn't valid there either (f(x2) > f(x3)), and find_minimum returns status -1 and NaN.

I pushed a change that sets v_mp and p_mp to exactly 0 wherever v_oc is 0, since 0 is the only voltage in [0, v_oc]. Those elements still get a placeholder bracket so scipy sees x1 < x2 < x3, but its result is discarded. The test now asserts v_mp == 0 exactly.

[Blizzeq]

Merged main and moved the whatsnew entry to v0.16.2, since v0.16.0 is out.