Support lambda-based parallel operations

[IvanGrigorik]

PyKokkos has a way to support the execution of lambda-based expressions execution strictly inside of workloads:

@pk.workload
class someTest:
    def __init__(self, threads: int, value: int):
        self.threads: int = threads
        self.value: int = value
        self.sum: int = 0

    @pk.main
    def run(self):
        self.sum = pk.parallel_reduce(
            "name", self.threads, lambda i, acc: acc + self.value
        )

However, since they will be deprecated soon, we need to port this into workunits or similar semantics, since they are not supported by pykokkos from non-pykokkos functions. So if you have consecutive code that uses parallel_* operation with a lambda function like that:

main():
    value: int = value
    sum: int = 0
    threads: int = 10
    sum = pk.parallel_reduce("name", threads, lambda i, acc: acc + value)

PyKokkos would be unable to translate and execute it, since it is not supported by the framework yet.

I suggest adding a lambda-based expression based on these semantics, since it looks intuitive, and this allows us to move from PyKokkos-specialized syntax to Pythonized syntax itself.