Fix parametrized generics for dataclasses [GRF-6059]

[bcalvert-graft]

Description

Currently, if you have a @dataclass in Sematic where one of the attributes is a parametrized Generic[T] class, Sematic will complain. For example, on graft/sematic/main

from dataclasses import dataclass
from typing import Generic, TypeVar
import sematic

JS = TypeVar("JS")

@dataclass
class A(Generic[JS]):
    val: JS

    def test(self, a: JS) -> JS:
        pass

@dataclass
class B(A[int]):
    def test(self, a: int) -> int:
        return a + self.val


@sematic.func
def gen(inp: B) -> int:
    return inp.test(1)

b = B(val=1)
sematic.SilentRunner().run(gen(b))

TypeError: Invalid input type for argument 'inp' when calling '__main__.gen'. Cannot cast B(val=1) to <class '__main__.B'>: Cannot cast field 'val' of B(val=1) to <class '__main__.B'>: Type 'JS' is not a valid Sematic type: Expected a Sematic-supported type here, but got: ~JS. Please refer to the Sematic docs about supported types: https://docs.sematic.dev/type-support/type-support#what-types-are-supported

The problem is that Sematic is directly pulling field.type which for classes A/B above, returns JS for val. Instead Sematic needs to correctly resolve the JS (or other TypeVars) against the arguments to the Generic class. This PR fixes this issue by introducing a resolve_type method to handle this scenario.

Testing

$ pytest sematic/types/types/tests/ sematic/utils/tests/test_types.py

Manual Testing

In [1]: from dataclasses import dataclass
   ...: from typing import Generic, TypeVar
   ...: import sematic
   ...: 
   ...: JS = TypeVar("JS")
   ...: 
   ...: @dataclass
   ...: class A(Generic[JS]):
   ...:     val: JS
   ...: 
   ...:     def test(self, a: JS) -> JS:
   ...:         pass
   ...: 
   ...: @dataclass
   ...: class B(A[int]):
   ...:     def test(self, a: int) -> int:
   ...:         return a + self.val
   ...: 
   ...: 
   ...: @sematic.func
   ...: def gen(inp: B) -> int:
   ...:     return inp.test(1)
   ...: 
   ...: b = B(val=1)
   ...: sematic.SilentRunner().run(gen(b))
Out[1]: 2

[mick-graft]

LGTM