I am pretty sure that Task 2 has a different correct solution.
It should be 2023 because that is the result of the process which identifies the "year within which the production of a given model was stopped". This would be the year within which the production of a model was last observed, not the subsequent one.
To put it differently: If I stop doing something within continuous time, but observation is limited to discrete intervals, then the moment that I stopped doing that thing has to be within the interval i was last observed doing that thing. There is the obvious exception that when I stopped doing that thing at the exact boundary between two intervals, then obviously the thing being done was stopped in the subsequent interval; but in continuous time, the probability of that happening is Almost Surely zero. We could only argue that this is the case for our example when (for e.g. tax reasons?) the production of the last car of the year for every model ends exactly at midnight, the 31st of December.
