Nodata values are affecting raster contour styling when using early resampling

[AlisterH]

What is the bug or the crash?

"Contours" raster styling with early resampling draws contours at the edge of nodata areas when zoomed out. This causes a dramatic decrease in performance (often rendering just seems to go on forever)

I believe this is a longstanding issue

Steps to reproduce the issue

See styling options in the screenshots below.
The light green on the right side of this view is outside the raster extent. On the left side it is nodata.
Rendering is OK at this zoom level:

Not OK zoomed slightly further out:

Along the edge of the nodata area the resampling calculation appears to be considering the NODATA value of -9999 as an actual value.

For another raster I have in front of me which has 0.1m resolution (rather than 1.0m) it is OK at 1:129, but not OK at 1:258.

Versions

QGIS version
3.40.1-Bratislava
QGIS code revision
989853a

Libraries
Qt version
5.15.13
Python version
3.12.7
GDAL/OGR version
3.9.3
PROJ version
9.5.0
EPSG Registry database version
v11.016 (2024-08-31)
GEOS version
3.13.0-CAPI-1.19.0
SQLite version
3.46.1
PDAL version
2.8.1
PostgreSQL client version
16.2
SpatiaLite version
5.1.0
QWT version
6.3.0
QScintilla2 version
2.14.1
OS version
Windows 11 Version 2009

Active Python plugins
civilplan
1.0
Equal_area_slope_QGIS_Plugin
0.1
firstaid
3.2.0
FlowEstimator
0.21
joinmultiplelines
Version 0.4.1
LayerBoard
1.0.1
LockCanvasZoom
1.0
MemoryLayerSaver
5.0.2
nominatim
1.4.5
pathfinder
version 0.5
plugin_reloader
0.17
precisioncursor4qgis-main
1.1.D
profiletool
4.2.6
QCopycanvas
0.7
quick_map_services
0.19.36
QWater
3.3.2
searchlayers
3.1.1
Serval
3.32.0
valuetool
3.0.19
workbench
0.0.4
db_manager
0.1.20
grassprovider
2.12.99
MetaSearch
0.3.6
processing
2.12.99
basemaps
1.0

Supported QGIS version

• I'm running a supported QGIS version according to the roadmap.

New profile

• I tried with a new QGIS profile

Additional context

Sample data sourced from https://data.linz.govt.nz/layer/106410-auckland-north-lidar-1m-dem-2016-2018/
92L2W-92KY1.zip

Perhaps this is could be a limitation of gdal rather than being the fault of QGIS itself...?