Fix Line failing with buff and path_arc (issue #4132)

[irvanalhaq9]

Overview: What does this pull request change?

Fixes issue: #4132

Motivation and Explanation: Why and how do your changes improve the library?

Issue Explanation

When buff is given, Line internally calls:

self.pointwise_become_partial(vmobject=self, a=buff_proportion, b=1 - buff_proportion)

(see: pointwise_become_partial in Line)

Since vmobject is self, modifying self.points also modifies vmobject.points.
This leads to a problem in the code (lines 1942-1949 and beyond),
where self.points = np.empty(...) also affects vmobject.points, erasing the data before it is used.

Solution

Before setting self.points = np.empty(...), we first make a copy of vmobject.points
to ensure we are working with valid data.

Reviewer Checklist

• The PR title is descriptive enough for the changelog, and the PR is labeled correctly
• If applicable: newly added non-private functions and classes have a docstring including a short summary and a PARAMETERS section
• If applicable: newly added functions and classes are tested

[irvanalhaq9]

Seeking feedback from reviewers:

This PR modifies the pointwise_become_partial method. I made this change after investigating and tracing the last working commit for Line with buff and path_arc. I found that it worked correctly in commit 3a714115, but after commit 374eeeba, it no longer worked. It appears that PR #3760 introduced this issue (#4132).

Since that PR modified pointwise_become_partial, my initial fix focused on this method by making a copy of vmobject.points before setting self.points = np.empty(...) to prevent unintended in-place modifications.

However, I am now considering an alternative approach: instead of modifying pointwise_become_partial, a simpler solution might be to change how Line internally calls it. Instead of passing vmobject=self, what if we pass vmobject=self.copy()?

What do you think about this approach?
If this seems like a better solution, I can update the PR accordingly.