load_sample: fix sample data registry for 7 datasets

[neutrinoceros]

PR Summary

This is complementary with yt-project/website#103 and should fix load_sample for 6 more datasets that were not following the dominant naming conventions in tarballs

this will also need to be followed by a systematic cleanup akin to #3317

[cphyc]

Why are the hashes changing?

[neutrinoceros]

Because the original tarballs didn't follow the de facto rule implemented for every other data sample: files need to unfold to a directory with the same name as the tarball (minus the .tar.gz extension), which also matches the "key" that's passed to load_sample

[neutrinoceros]

this will very likely be superseded by #3366
Switching to draft for now, but I except this will end up being closed rather than merged.

[matthewturk]

@neutrinoceros should this be closed?

[neutrinoceros]

Probably, but I'd like to keep it around for now if you don't mind.

[neutrinoceros]

closing this to clear the backlog as much as possible.
#3366 is the only change we need, and the one performed here can be done without a PR when #3366 is completed