ENH: Allow npv calculation to be broadcastable

[Kai-Striega]

This PR applies broadcasting operations to npv calculations. This is significantly faster than iterating over an array of cashflows as is shown by the below code snippet.

import numpy as np
from numpy.testing import assert_allclose
import numpy_financial as npf

import time


def faster_npv(rate: float, values: np.ndarray):
    """ A faster way to calculate NPV for several projects based
    on numpy arrays. """
    values = np.atleast_2d(values)
    timestep_array = np.arange(0, values.shape[1])
    npv = (values / (1 + rate) ** timestep_array).sum(axis=1)
    try:
        # If size of array is one, return scalar
        return npv.item()
    except ValueError:
        # Otherwise, return entire array
        return npv


if __name__ == "__main__":
    rate = 0.05
    no_simulations = int(1e6)
    lifetime = 10  # years

    capex = np.random.normal(loc=1000, scale=100, size=no_simulations)
    opex = np.random.normal(loc=100, scale=10, size=no_simulations)
    values = np.array([capex] + [opex] * lifetime).transpose()

    start_time = time.time()
    all_npv_slow = [npf.npv(rate, values[i]) for i in range(no_simulations)]
    print(f"Standard NPF calculations took {time.time() - start_time}s.")

    start_time = time.time()
    all_npv_fast = faster_npv(rate, values)
    print(f"Faster NPV calculations took {time.time() - start_time}s.")

    assert_allclose(all_npv_fast, all_npv_slow)

closes #52

[Kai-Striega]

pinging @jowahlin does this look good to you?