[FEATURE] Provide more granular callback mechanism

[TomeHirata]

Feature Area

Other (please specify in additional context)

Is your feature request related to a an existing bug? Please link it here.

N/A

Describe the solution you'd like

The current implementation for the callback mechanism of CrewAI is enabled by passing a callback function as the Task attribute or Agent attribute. However, since the callback function is called only after the execution, it is challenging to have full observability including latency of the execution. Therefore, it would be great if a new callback attribute is added to Task, Agent and so on that is called before the execution.

def pre_method_callback(input):
  ...

def post_method_callback(output):
  ...

from crewai import Agent
agent = Agent(
    role="Senior Data Scientist",
    goal="Analyze and interpret complex datasets to provide actionable insights",
    backstory="With over 10 years of experience in data science and machine learning, "
              "you excel at finding patterns in complex datasets.",
    llm="gpt-4",
    pre_callback=pre_method_callback,
    post_callback=post_method_callback,
)

Describe alternatives you've considered

Alternatively, the callback function could be configured as a global configuration as DSPy does. In this approach, one callback class will be defined and configured globally. The callback class implements the callback methods for each major class (e.g. Crew, Task, Agent) separately.
https://dspy.ai/tutorials/observability/

Additional context

No response

Willingness to Contribute

Yes, I'd be happy to submit a pull request