Align _AbstractRenderer.tube with the PyVista implementation

[karpovantonme]

Reference issue (if any)

None.

What does this implement/fix?

_AbstractRenderer.tube declares

def tube(self, origin, destination, radius=0.001, color="white", scalars=None,
         vmin=None, vmax=None, colormap="RdBu", normalized_colormap=False,
         reverse_lut=False):

but _PyVistaRenderer.tube, the only implementation, ends with opacity=None and uses it (_add_mesh(..., opacity=opacity)), and _plot_sensors_3d calls it that way when drawing the fNIRS source-detector pairs:

actor, _ = renderer.tube(
    origin=sources[ch_name][np.newaxis] * unit_scalar,
    destination=detectors[ch_name][np.newaxis] * unit_scalar,
    radius=0.001 * unit_scalar,
    opacity=sensor_alpha["fnirs_pairs"],
)

So the parameter that the live caller relies on is missing from the contract. This adds opacity=None to the abstract signature, in the same position as the implementation.

Two things in the same docstring were also off:

• backface_culling is documented but is not a parameter of either signature. It is a real parameter of mesh and surface, just not of tube, so the block looks copied. Removed.
• normalized_colormap is in both signatures and was not documented. Added, reusing the wording already used for it in contour in the same file.

The abstract method body is pass, so there is no runtime change. No changelog entry, since nothing user-facing changes — happy to add one if you would rather have it recorded.

While reading around this I noticed whiten_evoked documents diag : bool (default False) while the signature has diag=None — None was introduced deliberately in ce2486c (#4601), but the body only does if diag: so both values behave the same and the docstring was never updated. That is a separate one-liner and I did not want to mix it in; say the word and I will send it.

AI assistance, per the policy: the mismatch was surfaced by a sweep comparing numpydoc Parameters blocks against actual signatures, run with Claude Opus in Claude Code, which also wrote the patch; it checked the implementations and callers of tube and the history of the lines. I read the diff and the call site before submitting.

This is my second PR here today (the other is #14124) and they are unrelated — apologies if that is more than you would like at once; I am happy to close either and resubmit later.

[welcome]

Hello! 👋 Thanks for opening your first pull request here! ❤️ We will try to get back to you soon. 🚴

[larsoner]

Seems like a reasonable fix... even if the Renderer interface isn't exactly meant for public use.

Could you add a little docs/changes/dev/14125.bugfix.rst for this? And if you can log in to CircleCI with your GitHub account, it should make the next push run properly

[karpovantonme]

Changelog entry added, along with the names.inc line.

CircleCI is sorted now, I am logged in. The last run predates that, so it still shows the pipeline never starting. Could you kick it off, or should I push again to trigger it?

[larsoner]

I started it. Marking for merge-when-green. Thanks in advance @karpovantonme !