Adopt SPEC0 about minimum supported dependencies

[niksirbi]

Is your feature request related to a problem? Please describe.
The scientific Python ecosystem has agreed on a set of mimum supported dependecies, which are detailed in ther SPEC0. We are mostly compliant to these, and it wouldn't take much work to make us fully compliant, and get a fancy badge. Since all we do depends heavily on "core" scientific Python libraries, it makes sense to move in tandem with them.

Describe the solution you'd like

• Drop support for Python 3.10
• Add support for Python 3.13
• Set minimum version requirements those dependencies that are part of the spec0 (e.g. numpy, pandas, xarray
• Add the badge to README

Considerations
How often should check and update these things? Could we somehow automate that?

[niksirbi]

CC @adamltyson

[adamltyson]

I think it can be hard to automate this, because we often have to wait for other dependencies to catch up.

I don't see a big issue with us lagging behind by a few months, it's unlikely to cause problems for users.

[niksirbi]

I did some trials and we have to wait for Python 3.13 support for a while.

On my silicon mac, numba, which is a requirement of xarray[accel], fails to instal on a Python 3.13 environment.
See numba/numba#9413

So you're absolutely right about lagging behind for a few months.

[niksirbi]

My WIP is in this branch, for later reference.