Op. 8 is a collection of pre-existing pieces put together most likely based on business decision by the publisher. First four concertos already existed before being published as part of this collection. Earlier hand written manuscript already collects these four concertos together. Later composer gives special group name for these four concertos by using Four Seasons as a title when dedicating op. 8 (first published version): “Pray do not be surprised if, among these few and feeble concertos, Your Most Illustrious Lordship finds the Four Seasons which have so long enjoyed the indulgence of Your Most Illustrious Lordship’s kind generosity”. Dedication also hints that count (dedicatee) had known manuscript copies before the work (or collection) was being published as part of op. 8.
We could argue about the definition of work. It all depends if we count collections as works. In MB we have commonly stored collections in work entities.
These four concertos are worldwide known as Four Seasons and different databases, libraries and stores typically include separate entry for this collection. For me it makes sense to store this on MB work entity when there’s separate VIAF code, Wikidata ID, IMSLP-page or basically separate entry on every service we know. It’s not only that we link to these services but many of them are linking to MB.
Composer himself did group these together and would feel silly to say “we know it better”. I see no harm done even if this might not be a work by all definitions. While academic scholars, printed scores, Encyclopædia Britannica, Wikipedia and hundreds of releases cite this being “the most famous work” of Vivaldi we could include it to our database by using the name composer himself gave to it.