Use the qtoolkit and jobflow PyPI packages, add direct pydantic dep

[ml-evs]

Closes #30

Related to #29

[ml-evs]

@gpetretto / @davidwaroquiers this would be helpful to merge before #29 is finished, then we can undo the changes after. (I also realise I do not have the power to request reviewers here... perhaps I could be added as a maintainer?)

[gpetretto]

The current jobflow[strict] version pins pydantic to version 2.4.2. The latest jobflow version supporting pydantic <2 was 0.1.13. Maybe it would be better to also pin the jobflow requirement to that version? I am not sure if having pydantic<2 here could cause some conflicts.

[davidwaroquiers]

Just the comment on jobflow version wrt pydantic version. Not sure about it.

[ml-evs]

The current jobflow[strict] version pins pydantic to version 2.4.2. The latest jobflow version supporting pydantic <2 was 0.1.13. Maybe it would be better to also pin the jobflow requirement to that version? I am not sure if having pydantic<2 here could cause some conflicts.

It could be pinned, but this seems like the lightest weight way of doing it -- all that #29 needs to do afterwards is lower pin pydantic~=2.0 and then jobflow>=0.1.14 will be used. Up to you if you'd rather pin jobflow too though!

[ml-evs]

I guess @gpetretto you also mean whether it is worth even adding pydantic as a dependency, instead just going along with jobflow? I think as we are using specific pydantic code here, then it should definitely appear as a dependency, even if the strict mode of jobflow is installing a specific version of it already. There were quite a few breaking pydantic changes even within the v1.x series...

[gpetretto]

If pip resolves the correct version of jobflow to use it is fine as it is.