Support running Evals on specific tasks

[avivex1000]

In the current implementation, evals is not easily accessible or runnable without downloading the source code.
The proposed implementation will allow importing the evals runner, defining the preferred models to evaluate against and will run the evaluations on a given task.

Suggested usage:

from declarai.evals import Evaluator

models = [
    Declarai(provider="openai", model="gpt-3.5-turbo"),
    Declarai(provider="openai", model="gpt-4")
]

def test_task() -> str:
    """
    say something
    """

evaluator = Evaluator(models=models)
evaluator.run(test_task)