fixes for Processing geometry checker algorithms

[alexbruy]

Description

Various fixes for Processing check/fix geometry algorithms:

• raise an exception in geometry checker algorithms if selected unique ID field contains duplicated values to prevent failures at the geometry repair step (fixes Check geometry (angle) does not check whether feature ID is unique #61398)
• avoid use of QgsProject::instance() in algorithms
• correctly mark algorithms which require project with the corresponding flag
• added cancellation support in geometry checks

[alexbruy]

@Djedouas what do you think?

[github-actions]

🪟 Windows Qt6 builds

Download Windows Qt6 builds of this PR for testing.
(Built from commit 4256f9d)

🍎 MacOS Qt6 builds

Download MacOS Qt6 builds of this PR for testing.
This installer is not signed, control+click > open the app to avoid the warning
(Built from commit 4256f9d)

[Djedouas]

Looks good to me, thanks!

[alexbruy]

@nyalldawson, @Djedouas I have updated PR with another approach. Now check for duplicated IDs is performed in the geometry checker when unique ID field index is set in the geometry checker context.

I also added support for cancellation as most of the checks do not have it.

@Djedouas may I ask why are you using QgsProject::instance() in the algorithms and not project from the context? Also I noticed that all algorithms are marked with NoThreading flag. According to you

And big layers are what we expect to have with these check algorithms.

But with checks/fixes running in the main thread they seem not very usable with large datasets.

[Djedouas]

Thank you for the support of the algorithm cancellation :)

I think that you are right to throw an exception if the processing is cancelled. But we need to homogenize with the cancellation check that is in most check algorithms during the for loop on check errors (i.e. https://github.com/qgis/QGIS/blob/master/src/analysis/processing/qgsalgorithmcheckgeometryarea.cpp#L195), and throw an exception as well.

Going through a rapid verification of check algorithms, there are some that are missing this cancellation check (i.e. https://github.com/qgis/QGIS/blob/master/src/analysis/processing/qgsalgorithmcheckgeometryselfintersection.cpp#L186 maybe it's the only one?).

Thank you as well for the unique field ID verification inside the geometry checker 👍🏼 looks good to me. That's a huge improvement.

@Djedouas may I ask why are you using QgsProject::instance() in the algorithms and not project from the context? Also I noticed that all algorithms are marked with NoThreading flag.

Good question :) At the beginning of this work to implement these algorithms, I surely chose the multithreaded option. But then there were a lot of segmentation faults that I was unsure where they were coming from. So to be sure they were not coming from the processing themselves, I put the NoThreading flag.

Your question makes me think that it could be because the processings are using QgsProject::instance() instead of the context project.

We have to end up with multithreaded algorithms to handle large datasets. If you manage to achieve this, It would be great!! (and I suggest adapting the PR title then 😉 )