Again, I want to reassure that nothing has changed on this endpoint and it has remained similar in behaviour since the Twitter Search API came online ~7 years ago.
The language is deliberately vague, as the algorithm and indexing has never been fully defined externally. To answer the specific question, some users and terms, specifically those created very recently or heavily flagged as potentially abusive or spammy, may be excluded from the limited 7 day index for standard search. Equally, at times of load some Tweets or users may be excluded from this index. It is not possible to describe every single condition under which Tweets may not be included.
The behaviour of the standard search endpoint has categorically not changed as a result of the premium API; the new endpoint provides net new functionality built on the enterprise platform.