You are already pretty close. But as you see Picard now sometimes has distributed your files over multiple similar versions of a release.
That’s happening because you are either using the scan function (which looks at each file individually) or have lookup on individual files. In both cases Picard looks up file by file and tries to figure out for each file onto which release it thinks this fits best.
If you are looking up entire albums you should instead first use the “Cluster” button on the files you have loaded to the left (Unclustered files). Picard then tries to group those files into files that seem to belong to the same release. Picard calls this generically a “cluster”, meaning files belonging somehow together.
In your example that should result in a single cluster “Rebel Yell” with all 9 tracks. Once you have clustered the files you can then use “Lookup” on that cluster. Picard then looks for a best fit for the entire group, meaning it will search for a release “Rebel Yell” with 9 tracks.
Depending on what you are looking for and what files you are tagging and how the current metadata is you might still need to do Scan or Lookup on individual files. In those cases Picard might find the songs on different releases, including compilations. But you can always check the results and drag files around. You can also configure Picard to prefer albums over compilations or singles in Options > Metadata > Peferred > Releases.
You can also search the MusicBrainz database directly and load any release manually into Picard.
See the docs at https://picard-docs.musicbrainz.org/en/usage/retrieve.html on the various ways to load releases into Picard.
I think it is the name othat confused you. “Promo Only: Platinum Series” is a compilation, and it just happens to include one of the songs you were looking up.
My suspicion is that you used “Scan” for this all. “Scan” does an acoustic analysis of the files and tries to find a match. But acoustically there is no difference between that song on the album or on a compilation.