Fix assertions and improve error messages when adding submobjects

[chopan050]

Closes #3755.

Initially this PR only improved 2 things:

1. When adding into a Mobject a value which is not a Mobject, or there's an attempt to add a mobject to itself, the error messages are now more descriptive, indicating:

   • what's the value being added
   • what's its type
   • at which index is the issue happening
   • into which Mobject you're attempting to add the value

   Example:

>>> img = ImageMobject("cat.jpg")
>>> img.add(Circle(), 3)
# TypeError: Only values of type Mobject can be added as submobjects of ImageMobject, but the value 3 (at index 1) is of type int.
>>> img.add(img)
# ValueError: Cannot add ImageMobject as a submobject of itself (at index 0).

2. Refactor the code by creating a Mobject._assert_valid_submobjects() method which makes this check, and calling it on every method which required this assertion, which are Mobject.add(), Mobject.insert() and Mobject.add_to_back(). Previously, those methods duplicated the code for checking the submobjects, but in different ways, with very different error messages, and occasionally not always checking the types of the submobjects.

But I noticed that, while VGroup overrided the add() method to be more restrictive (by checking that the submobjects are all instances of VMobject), the VMobject class itself didn't do that, which causes the issue #3755: you could add a Mobject to a VMobject such as Circle, and a Mobject could be part of the family of a VGroup without being a direct child of it.

Plus, the other methods insert() and add_to_back() weren't overridden, so one could still do this:

>>> VGroup().add_to_back(Mobject())
VGroup(Mobject)

So, this PR also fixes that.

3. Any VMobject, not only VGroups, now asserts on add(), insert() and add_to_back() that all the submobjects are instances of VMobject. To do that, it suffices to override the _assert_valid_submobjects() method in VMobject. It is not necessary to actually touch the add(), insert() and add_to_back() methods. Because the VGroup.add() docstrings have an useful example, I decided to leave that override in the code, even if it now contains only a return super().add(*vmobjects) line.

>>> Circle().add(Mobject())
# TypeError: Only values of type VMobject can be added as submobjects of Circle, but the value Mobject (at index 0) is of type Mobject. You can try adding this value into a Group instead.

Finally:

4. Repeat all the previous steps for OpenGLMobject and OpenGLVMobject to avoid breaking tests, and for more consistency.
5. Update the tests to take into account the more descriptive error messages.

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

[JasonGrace2282]

Changes look good to me!
Think you missed one or two places though
https://github.com/chopan050/manim/blob/0046b8496d1736ecf65386a719ba7a6971eb5ac3/manim/mobject/opengl/opengl_vectorized_mobject.py#L1691-L1692

https://github.com/chopan050/manim/blob/0046b8496d1736ecf65386a719ba7a6971eb5ac3/manim/mobject/opengl/opengl_vectorized_mobject.py#L1803-L1804

[chopan050]

Thanks for highlighting those places! I fixed them, as well as some other places I missed.