[Feature Request] Support for recovering NumPy arrays (`numpy.ndarray`)

[liblaf]

First of all, thank you for creating coredumpy! It's an useful tool for debugging, and I've found it convenient for inspecting torch.Tensor objects from core dumps.

My feature request is to add similar support for NumPy arrays (numpy.ndarray). Being able to inspect numpy.ndarray objects directly from a core dump would significantly enhance the debugging capabilities.

Sample Python script:

import coredumpy
import numpy as np

coredumpy.patch_except()


def main() -> None:
    arr: np.ndarray = np.ones((3, 3))
    raise RuntimeError
    print(arr)


if __name__ == "__main__":
    main()