[3.13] gh-122478: Remove internal frames from tracebacks in REPL (GH-122528)

[cfbolz]

Attempt to redo the missing backport of gh-122528, superseding #122817 (which is conflicting now). /cc @pablogsal

Frames of methods in code and codeop modules was show with non-default sys.excepthook.

Save correct tracebacks in sys.last_traceback and update traceback attribute of sys.last_value and sys.last_exc.
(cherry picked from commit e73e7a7)

• Issue: The new REPL outputs different tracebeck when custom sys.excepthook is used #122478

[serhiy-storchaka]

What is the difference with #122817?

[cfbolz]

What is the difference with #122817?

oops, there isn't one. #122817 had a conflict and I didn't know the right github approach to fix the conflicts of a miss-islington backport PR, so I opened this new one instead. I'll close #122817.

[serhiy-storchaka]

You can use gh pr checkout to get the PR code of other user, change it, commit, and push back. Very handy.

There are some differences between #122817 and this PR, but I do not understand them (there is nothing similar in the main and 3.12 branches). This is why I asked.

[cfbolz]

You can use gh pr checkout to get the PR code of other user, change it, commit, and push back. Very handy.

thank you!

There are some differences between #122817 and this PR, but I do not understand them (there is nothing similar in the main and 3.12 branches). This is why I asked.

Yes, it took me ages yesterday to find out what happened. Basically the pyrepl PRs were (manually?) backported out of order, and this one was one of the skipped ones. this led to conflicts in the automatic backports, and it's the reason why this PR is different than #122817. but I think we're almost back to a consistent state.