Even Rouault

Rebased on top of lastest master to take into account https://github.com/OSGeo/PROJ/pull/4098 ("Make NKG Transformation names unique")

@kbevers the consistency checks detected 2 remaining occurences of duplicated NKG transformation names, which I propose to resolve with a8db330c09734d248726c15d421af21825ade1af

GDAL fix queued in https://github.com/OSGeo/gdal/pull/11248. I've confirmed it fixes tests/test_merge.py::test_data_complex Thanks @schwehr for bisecting to the offending PR

> I didn't add a progress callback because > 90% of the time was inside GEOS. thanks for the feedback! I've just readjusted the progress ratio based on that. Yes...

@dbaston Do you believe the GEOS team would welcome enhancements to add progress report and cancellation for this algorithm? (to be checked if that's possible in that case). I don't...

https://github.com/OSGeo/gdal/pull/10721 has been merged. Please rebase on top of latest master

continued in https://github.com/OSGeo/gdal/pull/11760

> why an open option instead of adding a new argument to GetLayer() or GetFeature()? OGRLayer::GetNextFeature() would definitely not be the appropriate entry point. This is way too late, as...

The downside of a dataset-level open option is for multi-layer datasets. This will require the JSON document to be something like: ```json { "layerA" : { }, "layerB" : { ...

... or have a OGRLayer::SetLayerDefn() (for symmetry with GetLayerDefn()) or OverrideLayerDefn() only implemented by drivers advertizing a OLCSetLayerDefn capability . This should be called *before* any feature has been retrieved...