There's no time frame I'd say. Buddy of mine is a mathematician for one of the game providers. Each game when released, to determine the RTP is tested through simulations while using virtual players, with different stakes, bonus buys etc. From that, pay tables are analysed .
I'd rather say that RTP is average over a specified number of game rounds, not on time period.
Other thing, volatility of the game makes the difference as well. From a single player point of view, if playing the game with 95% - high volatility, hit rate would be lower, but chance of winning bigger amount will be better, where in low volatile games (same RTP) I will win more often, but less.
I can't find now, there were some statistics per game, with hit rates on each slot (official info from providers), chances of hitting max wins per game etc.
If I remember well, winning max win on some of the Pragmatic Play games was around 1 in a million. Depends on the game.
Could be worth as well to check the licensing bodies (at the end of the day, games have to be approved by each license), so UKGC for example, requires games to be tested and RTP monitored to calculate the actual RTP, and compared to the expected (advertised) RTP. Each license will be different though.
Just an example from UKGC:
https://www.gamblingcommission.gov.uk/strategy/testing-strategy-for-compliance-with-remote-gambling-and-software-technical/5-live-rtp-monitoring#ref-11
Automatic translation: