At the moment the clang-tidy CI fails on each file with the following
message:
```
error: PCH file uses an older PCH format that is no longer supported
[clang-diagnostic-error]
```
This is because clang-tidy version 18 is used while QGIS
compilation is achieved with clang 15. Indeed, QGIS is built with
clazy support which relies on clang 15.
This issue is fixed by using clang version 15 instead of version 18 in
`docker-qgis-clangtidy.sh`.
* Handle when /collections endpoint returns paginated results
Servers usually return all collections in a single page.
If the results are paginated, get the 5 first pages, just in case the total number of pages is huge.
If 5 pages are still not enough, more will be fetched when scrolling the Collections list view.
* Reset collections page counter when connecting
When a new 3d is created
`Qgs3DMapSettings::configureTerrainFromProject` is called to configure
the terrain. This method calls `Qgs3DMapSettings::setTerrainSettings`
which sets the terrain settings and creates the terrain generator if
necessary. This generator is not updated if the settings have not
changed.
However, on a new 3d view, the terrain settings (`mTerrainSettings`)
have already been created with the default values and the
`QgsAbstractTerrainSettings` parameter of
`Qgs3DMapSettings::setTerrainSettings` also contains the default
settings. Therefore, this call does nothing and the terrain generator
is not created. This results in a 3d view which does not contain any
terrain.
This issue is fixed by also checking if the terrain generator has
already been created. If that's not the case, the terrain generator is
now created.
The flatTerrain test is updated to test this workflow.
Fixes: 8af23b10024be91a8e97a81832907393544a7fd5
Fix#53149
...and possibly #60625
The issue is deep into QT internals: the abstract spinbox
has a line edit wich holds the different datetime components
divided into "Sections", when the user selects all (e.g. with
CTRL+A) and then pastes a datetime string from the clipboard,
it may get pasted in the wrong position (i.e. not at the beginning
of the selection) because the internal position of the cursor
wasn't updated).
I tried hard to write a test (even tried QTest::click* and events
but I couldn't find a way to reproduce the issue programmatically
using the QT public API).
