More programmatic access to error locations in the traceback module

Feature or enhancement

I'd like for the traceback module to have a more programmatic way to access the fine-grained error locations that are printed in tracebacks starting in Python 3.11 (PEP 657).

Pitch

The new fine-grained error information in tracebacks is nice. But I'd like to be able to write a custom traceback printer that uses this information and prints it in a different way (I'd like to highlight the characters in the lines of source rather than having ASCII characters under them).

The logic for getting the locations of the printed fine-grained information is present in the traceback module, but unfortunately, it's only present inside of the format_frame_summary() method, which returns a string. That means that to access the information, I have to parse the string.

It would be convenient if the logic in that method were separated out into a separate method that returned the column numbers, so that it could be called from a superclass. I would be fine if this method is private, so that it wouldn't strictly speaking add a new API or hard-code the existing traceback information into the API.

Previous discussion

I haven't followed the discussions on PEP 657 so I don't know if this has been discussed previously. I don't see any mention of the traceback module in the PEP.

Start in Lib/traceback.py at format_frame_summary(), where the existing fine-grained traceback location logic is described. Determine how that logic could be separated to expose column numbers for custom traceback formatting without parsing the returned string. Done means callers can access the locations programmatically while preserving the existing traceback output.