No, everyone use the highest quality they can find.
Bulk down-sampling the lot is quick and easy, up-sampling the lot is impossible.
Yes, I guess everything should be highest quality, and then the devs can easily re-sample everything to 44.1 kHz (Trem's "High Quality" sound setting) and then further down-sample select sounds to 22 kHz to be smaller.
Just for future reference, you can quickly batch-convert between mono and stereo too, right?
SoX can convert stereo to mono by specifying to use 1 audio channel and to average the stereo channels:
$ sox stereo.wav -c 1 mono.wav avg
...and also convert mono to stereo by specifying 2 audio channels (-c 2):
$ sox mono.wav -c 2 stereo.wav
Audacity can process stereo to mono in batch using the StereoToMono command chain:
File -> Edit Chains... -> Add -> Give the chain a name like "Stereo to Mono" -> OK -> Insert -> double-click 'StereoToMono' -> OK -> Insert -> double-click ExportWav -> OK -> OK.
Then click File -> Apply Chain... -> Select your new chain -> Apply to files... -> Select the files you want to process -> OK.Audacity then makes a new folder named 'Cleaned' containing your new samples.
The downside of this method is that it only does files and won't let you recursively process folders.