Fix stacked area hide show bug

[emilykl]

Closes #7660

Fixes issue where some traces in stacked area charts were not rendering their fill correctly after certain sequences of hide/show operations in the legend.

Steps to test

Reproduce existing issue

1. Go to this Codepen (from the docs)
2. Follow the exact reproduction steps described in 7660:

• Double-click "trace 1" to show "trace 1" only
• Click "trace 1" again
• Now plot is empty
• Double click any trace to show all traces again

3. The orange trace should be missing a fill. This is the bug.

Confirm that this PR fixes the issue

1. In the Codepen, replace the plotly.js script src with the build for this branch from the CI: https://output.circle-artifacts.com/output/job/c4caeb45-0a0b-4e23-97b6-ebb26b8561c8/artifacts/0/dist/plotly.js
2. Repeat the steps under (2) above
3. All trace fills should render normally.

[camdecoster]

Since you're making _ownFill and _nextFill null, the if checks will always fail. So, let's delete them.

Suggested change

	// Clear any existing fill element references to ensure clean state
	// This prevents stale references when fill configuration changes
	trace._ownFill = null;
	trace._nextFill = null;
	var fillData = [];
	if(trace._ownfill) fillData.push('_ownFill');
	if(trace._nexttrace) fillData.push('_nextFill');
	// Clear any existing fill element references to ensure clean state
	// This prevents stale references when fill configuration changes
	trace._ownFill = null;
	trace._nextFill = null;
	var fillData = [];

[emilykl]

@camdecoster Tragically, _ownFill and _ownfill, as well as _nextFill and _nexttrace, are separate variables. 🙃

[camdecoster]

Ah, fair enough. I have words about this, but I won't include them here publicly.

[archmoj]

Thank you @emilykl for the fix.
The PR is missing a draftlog.
Other than that it looks good 💃

[emilykl]

@camdecoster I feel more confident about this fix now.

After reviewing the code, I'm satisfied that trace._ownFill and trace._nextFill are only intended to be used as temporary variables within the context of the createFills() function, and it's safe to reset them to null at the start of the function.

I think that lines 94-96 in the "before" code were intended to reset these to null as part of the D3 .exit() call; but I assume the bug manifested because .exit() is not always called on the fillJoin object, perhaps when the trace is not visible.