[cuegui] feat: Add job node graph plugin v2

[lithorus]

Link the Issue(s) this Pull Request is related to.
#888 (original PR)

Summarize your change.
This is an adapted PR to support QtPy and PySide6

It depends a fork of NodeGraphQt that has been adapted to use QtPy instead of Qt directly.

[lithorus]

It's pulling in a specific fork of NodeGraphQt which is not optimal. The question is wether or not it's better to just import the code from the fork instead of referencing an external repo (without releases).
It's licensed under MIT, so it should be ok?

[lithorus]

Obviously it fails the python 2 tests, but the PySide6 tests fails because of missing git command in the opencue:2019 image (due to the NodeGraphQt python module directly from a git repo).

[lithorus]

After I imported and modified the NodeGraphQt package I got it to pass all tests (including a simple one for the new graph), however the python2 ones doesn't test the GUI, so not sure if it will still work with python2.

Have only tested locally with PySide6 and not sure if it will work with PySide2.

[DiegoTavares]

After I imported and modified the NodeGraphQt package I got it to pass all tests (including a simple one for the new graph), however the python2 ones doesn't test the GUI, so not sure if it will still work with python2.

Have only tested locally with PySide6 and not sure if it will work with PySide2.

I wouldn't worry about python2 now, we're un"officially" not supporting it anymore.

[DiegoTavares]

It's pulling in a specific fork of NodeGraphQt which is not optimal. The question is wether or not it's better to just import the code from the fork instead of referencing an external repo (without releases). It's licensed under MIT, so it should be ok?

Commiting the fork as part of this project is not the right approach IMO, as it would prevent us from keeping thing in sync. I see two options to make this dependency maintainable.

1. Bring the NodeGraphQt fork in as a git submodule
2. Make downloading the fork part of the CI/CD pipeline

Although git submodules are a pain to maintain, I still think 1 is the best option, at least while there's no official release available.

[DiegoTavares]

Bring NodeGraphQt fork as a submodule.

[lithorus]

Looks like there has also been some updates to the pyside6 fork, will test with the latest changes to see if it might work

[lithorus]

A did a quick test and atleast with PySide6 it seems to work with just the updated fork. Will need to test it with PySide2 to see if it still requires changes.

[lithorus]

I had to remove the pyside6 test and upgrade python linting to CY2023 (from CY2022) to get all the tests to complete.
The changes also need changes to the documentation on how to clone with submodules

[DiegoTavares]

I'm planning to test this feature this week in the SPI environment to confirm we're able to bring the submodule and build it locally. If this works without any issues, I'm okay in merging this in the way it is. If I find it will be a pain to carry this dependency to our environment where downloading things from github is blocked at the network level, we might need to add changes to make this dependency optional.

[lithorus]

I have published a fork of NodeGraphQt as NodeGraphQytPy on pypi, which is using QtPy instead of Qt and have been using that instead of NodeGraphQt