[Maintenance] Updating the dependency versions and migration to `pyproject.toml`

[phoeenniixx]

Is your feature request related to a problem? Please describe.
Right now, pytorch-tabular mainly uses setup.py and some dependency versions haven’t been updated in a while. For example:

• python version is still from v3.8, which is now end-of-life.
• numpy versions do not yet support numpy 2.x.

Since many related projects (eg sktime, pytorch-forecasting) have already moved to pyproject.toml build system, I wanted to ask if we should do the same here?

Additionally, should we also update the linting and bring to closer to other sister packages (like ptf)?

Describe the solution you'd like
We should migrate to pyproject.toml, update all the outdated versions and update the linting.

FYI @manujosephv, @fkiraly

[phoeenniixx]

I was also looking at the workflows, there we are mainly using ubuntu, should we also add windows and macOS images there, in addition to updates to python versions?

[manujosephv]

I completely agree. We should modernize the package and bring it to the same level as other sister packages. We can also move to newer python versions and deprecate older end of life ones.

In general any changes which doesn't really affect the functionality directly, you can assume I'm on board with. 😊