Mesh Layer Rendering with Updated Canvas Fix

[JanCaha]

Description

There was a regression in Mesh Layer rendering for scalar datasets in case of "Updated Canvas" extent. This PR fixes it and also fixes fixes issue with style setting panel that could cause QGIS crash (test case was for this was added).

cc @uclaros

[github-actions]

🪟 Windows builds

Download Windows builds of this PR for testing.
Debug symbols for this build are available here.
(Built from commit 61e7229)

🪟 Windows Qt6 builds

Download Windows Qt6 builds of this PR for testing.
(Built from commit 61e7229)

[nyalldawson]

Where is this logic applied now?

[JanCaha]

It is applied in step of renderer creation. Basically, this part of the logic was applied twice, which caused issue (double rendering).

There is significant difference between mesh and raster renderer - mesh renderer is not stored at all, only the settings and mesh renderer is always created when it is needed. Thus there is no refresh of it.

I am not 100% sure that I properly understand the approach that was introduced with QgsRenderedLayerStatistics. Should the layer statistics be only applied on the renderer from this part of the code? And if rendered from anywhere else, just stick with data that are already set in mesh render settings?

[JanCaha]

@uclaros

I updated this to follow the use of QgsRasterLayerRenderer as closely as possible.

On creation of the renderer the QgsRenderedLayerStatistics is calculated but renderer settings are not updated with min max values. The update of renderer is only done from handleRenderedLayerStatistics() in QgisApp. This called from QgisApp::onRenderComplete(), which means that two rendering passes happen - first with old min max settings and after that second one, that is triggered from handleRenderedLayerStatistics(). This exactly follows steps for rendering Raster Layer.

This should fix the rendering pipeline, to be correct.