Push required python version to 3.10

[bizfsc]

As time goes by, I think it is time to push the minimal required python version once more. Why? Mainly new features for type hinting and code quality. Also 3.9 is already EOL and 3.10 is due in Oct 2026.

Actual benefits:

• Drop typing.Union / typing.Optional and use native hinting
• Structural Pattern Matching with match (PEP636): The library is full of if/elif chains dispatching on integer command bytes, OD object types, and NMT states. match makes this explicit, exhaustive, and readable
• TypeAlias for Explicit Type Aliases (PEP 613): Unambiguous and readable type aliases

[acolomb]

This is something I wanted to get started on, but haven't found the time yet. Certainly the right path forward, on all three points. But please make sure the changes re split up in smaller chunks, no "fix-all-the-things" commits that are hard to review and act as barriers in the Git history (for example when bisecting / blaming).

Especially for the Union and Optional fixes, I think there are more important issues to work on, so we can transition to the new style gradually.

I was pondering about a final 3.9 compatible release with the accumulated changes so far, before we take this dive.

[bizfsc]

This is something I wanted to get started on, but haven't found the time yet. Certainly the right path forward, on all three points. But please make sure the changes re split up in smaller chunks, no "fix-all-the-things" commits that are hard to review and act as barriers in the Git history (for example when bisecting / blaming).

yes! totally agree, the one trivial changes PR (#647) was an exception, because there I didn't see the benefit splitting this up even more in like three or four separate PRs.

I was pondering about a final 3.9 compatible release with the accumulated changes so far, before we take this dive.

I think this is a very good idea. No need to rush things