gh-71679: Enhance bytearray_repr with bytes_repr's logic

[furkanonder]

• Issue: Enhance bytearray_repr with bytes_repr's logic #71679

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @sunmy2019 for commit bc0270d 🤖

If you want to schedule another build, you need to add the 🔨 test-with-refleak-buildbots label again.

[sunmy2019]

Target to 3.13

[sunmy2019]

I think there are some problems here.

1. We cannot utilize the PyBytes_Repr from Python side since bytes.__repr__ will do checks.

>>> bytes.__repr__(bytearray([1]))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: descriptor '__repr__' requires a 'bytes' object but received a 'bytearray'

2. Although PyBytes_Repr is a public API, applying it for not bytes (but supporting buffer protocol) is still weird. As it clearly states, it's a repr of bytes.

3. Using the PyObject_GetBuffer for bytes slightly hurts the performance of bytes objects.

I think 1 and 2 is the main problem here. I would propose a new function that starts with PyBuffer_*, maybe PyBuffer_Repr.

Then, move the implementation to some function foo.

Then, call this foo from both PyBuffer_Repr and PyBytes_Repr.

[sunmy2019]

@serhiy-storchaka What do you think?

[serhiy-storchaka]

This PR contains only the part of the original patch. It does not change the repr of bytearray.

And it shares the flaw of the original patch -- it does not work with nested classes and classes with non-ASCII names.

See #138181 that solves this problem in other way.