Celery .delay() vs .apply_async() re MyPy

celery.mypy.py

from celery import Celery
from typing import Any

app = Celery('tasks', broker='pyamqp://guest@localhost//')

@app.task
def add(x: int, y: int) -> int:
    return x + y

# Usage examples
result = add.delay(1, 2)  # This will pass mypy validation
result = add.apply_async(args=(1, 2))  # This will also pass

# These will raise mypy errors
result = add.delay("1", 2)  # Error: Argument 1 to "delay" of "TaskType" has incompatible type "str"; expected "int"
result = add.delay(1, "2")  # Error: Argument 2 to "delay" of "TaskType" has incompatible type "str"; expected "int"
result = add.apply_async(args=(1, "2"))  # Error: Argument "args" to "apply_async" of "TaskType" has incompatible type "Tuple[int, str]"; expected "Tuple[int, int]"

# This will pass mypy but fail at runtime
result = add.apply_async(kwargs={"x": 1, "y": 2, "z": 3})  # mypy doesn't check kwargs keys