types.Union objects can contain reference cycles, therefore causing memory leaks.

E.g.::

```
import sys, gc
from typing import TypeVar

gc.collect()
for _ in range(10):
 sys.gettotalrefcount()
 T = TypeVar('T')
 U = int | list[T]
 T.blah = U
 del T
 del U
 gc.collect()
```
Result:
84470
0
84488
0
84504
0
84520
0
84536
0

I'm sending a small PR soon to implement GC methods to fix this.

New changeset 1097384ce964dd63686b1aac706cd0fa764c2dc9 by Ken Jin in branch 'main':
bpo-44553 : Implement GC methods for types.Union (GH-26993)
https://github.com/python/cpython/commit/1097384ce964dd63686b1aac706cd0fa764c2dc9

New changeset 0856134542c1067d02c9027ddbb2a61916907e92 by Miss Islington (bot) in branch '3.10':
bpo-44553 : Implement GC methods for types.Union (GH-26993)
https://github.com/python/cpython/commit/0856134542c1067d02c9027ddbb2a61916907e92

Thanks for the helpful review and merge Serhiy.

This commits seems to have broken the tracerefs buildbots:

https://buildbot.python.org/all/#/builders/484/builds/322
https://buildbot.python.org/all/#/builders/678/builds/127

Please, take a look at soon as possible, otherwise per the buildbot policy we will need to revert the commits to avoid masking other issues.

New changeset bc3961485639cc73de7c4c7eed1b56f3c74939bf by Pablo Galindo in branch 'main':
bpo-44553: Correct failure in tp_new for the union object (GH-27008)
https://github.com/python/cpython/commit/bc3961485639cc73de7c4c7eed1b56f3c74939bf

New changeset 000b9e803a7ec067da0a43f9a3fec16f1078215a by Miss Islington (bot) in branch '3.10':
bpo-44553: Correct failure in tp_new for the union object (GH-27008) (GH-27009)
https://github.com/python/cpython/commit/000b9e803a7ec067da0a43f9a3fec16f1078215a

Thanks for the fix, Pablo!

Ken Jin, we learned something today!

Woops, thanks for the quick fix Pablo. I'm guessing the takeaway here is
that we should start tracking a GC-ed object as early as possible?

Oh I didn't see Pablo's message on the PR:

> @Fidget-Spinner For next occasions, the problem with this is that PyObject_GC_Del cannot be called like that because it overrides a bunch of cleanups like the call to _Py_ForgetReference and other possible handling.

Thanks for the explanation!