Pickling numpy.busdaycalendar failed when using Multiprocessing in Python

Question

I'm trying to multiprocess a function whose parameters contain a numpy.busdaycalendar object. It gives me this error:

TypeError: cannot pickle 'numpy.busdaycalendar' object

Below is a simple code to replicate the issue:

from concurrent.futures import ProcessPoolExecutor
import numpy as np


def shift_one_day(x,constant):
    return np.busday_offset(x,1,busdaycal=constant)


def main():
    constant = np.busdaycalendar()

    
    variables = [np.datetime64('2023-03-21'),np.datetime64('2023-03-22'),np.datetime64('2023-03-23')]

    with ProcessPoolExecutor() as executor:
        futures_list = []
        for v in variables:
            futures_list.append(executor.submit(shift_one_day,v,constant))
        result_list = [future.result() for future in futures_list]
    print(result_list)


if __name__ == '__main__':
    main()

My question is:

1. Is there a way to make the busdaycalendar object pickle-able while maintaining its functionality?
2. If not, is there any workaround in this situation? Side information is that in all the function calls of shift_one_day, the constant object are all the same and doesn't need changing.

I'm using windows 10 so it seems no "fork" option for multiprocessing is available. Environment: python=3.12.2 numpy=1.26.4. Packages are from conda-forge.