ctx.download_resource creates directories needlessly
When target_path is None, the download_resource function is expected to download the resource into a temporary file.
What it actually ends up doing, is downloading the resource into a temporary file located in a brand new temporary directory:
Way too many directories created for no reason
In order to do proper cleanup, the caller needs to know that they should delete the temporary file's directory, and not just the temporary file itself
(Leaving this issue here rather than deleting it, so others who may wonder about this will find the explanation)
When target_path is None, we download the resource by preserving its basename. That requires us to download into a new directory each time.
Overall the best approach is to always provide target_path, and then delete the resource after usage.