Inconsistent behavior with Literals depending on the amount of elements it contains

This is an issue I ran into working on #9097.
Code adapted from this test.

Literals with one value:

from typing import List
from typing_extensions import Literal

lit1: Literal[1]
lit2: Literal[2]

arr2 = [lit1, lit2]

reveal_type(arr2)

➜  mypy git:(master) python3 -m mypy leba_misc/test2.py
leba_misc/test2.py:9: note: Revealed type is 'builtins.list[builtins.int*]'

Literals with multiple values:

from typing import List
from typing_extensions import Literal

lit1: Literal[1]
lit2: Literal[2, 3]

arr2 = [lit1, lit2]

reveal_type(arr2)

➜  mypy git:(master) python3 -m mypy leba_misc/test2.py
leba_misc/test2.py:9: note: Revealed type is 'builtins.list[Union[Literal[1], Literal[2], Literal[3]]]'

On Python 3.7 and mypy 0.790.

Inizia eseguendo con mypy i due snippet di riproduzione dell’issue e confrontali con test-data/unit/check-literal.test intorno al caso collegato. Traccia il comportamento di revealed-type nell’inferenza di liste che coinvolge valori Literal, quindi aggiungi o aggiorna un test di regressione in modo che i casi con un valore e con più valori si comportino in modo coerente.