gh-122996: Clarify encoding behavior in dbm module documentation

[furkanonder]

• Issue: the dbm docs are vague about what encoding is used when a str is stored #122996

---

📚 Documentation preview 📚: https://cpython-previews--138030.org.readthedocs.build/

[serhiy-storchaka]

I thought UTF-8 is always used. What backends use different encoding?

[furkanonder]

I thought UTF-8 is always used. What backends use different encoding?

Yes, UTF-8 is the default in most cases. But in this section, I wanted to reference SQLite databases specifically. SQLite can use different encodings. But for the sake of simplicity, I didn't mention SQLite directly in the documentation.

[serhiy-storchaka]

If this is so, then I think that this is a bug in the SQLite backend. The main data type is binary, setting string key and value is acceptable (for compatibility with Python 2), but you get bytes objects when read them back, and their values should be predicable and backend-independent.

I think we can live with a difference in handling surrogate characters.

cc @erlend-aasland