CASSANDRA-17047(trunk): Fix queries for DROP column when schema propagation takes time

[blerer]

[blerer]

Regarding the growing complexity of this class, perhaps a bit of unsquahsing would be good. I mean consider putting the logic for each Cassandra version range in a distinct block - for example https://github.com/apache/cassandra/pull/1286/files#diff-a97d038e026a08e915c6a2e0f736ffc9041e11c2fc0d0c13ee82d317cc88c435R1073 - I know it would make the code longer but it would be easier to follow.

No sure that it will really improve things.

One more thing is avoiding negations in names - I particularly mean hasNoExtra... variables. Perhaps it is only me, but I find it easier to read !hasExtra than hasNoExtra because the negation is put in the beginning :)

It might be my dyslexia but for me it is the opposite. :-) I also found slightly weird to compute a value and always check its negation.