Scipy 1.9.x中Beta-Binomial分布拟合异常：是否存在破坏性变更？

将Scipy版本从1.8.1升级至1.9.0，保持Numpy版本为1.24.4后，项目运行出现异常。以下代码可复现该问题，场景为使用Scipy的Beta-Binomial分布拟合概率分布：

from collections import Counter

import numpy as np
from scipy import stats
from scipy.optimize import differential_evolution


def func_nll(free_params, *args):
    """Negative log likelihood used in the optimizer to fit the best probability distribution available.
    More information on the optimizer used in
    https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.differential_evolution.html

    Arguments:
        free_params: probability distribution parameters under tuning. Depending on the probability
        distribution to fit, there can be one or more 
        *args: custom distribution arguments, including the 

    Returns:
        Negative log likelihood value used in the optimizer
    """
    dist, x, shift = args
    # negative log-likelihood function
    negativeLogLikelihood = -dist.logpmf(x, loc=shift, *free_params).sum()
    if np.isnan(negativeLogLikelihood):  # occurs when x is outside of support
        negativeLogLikelihood = np.inf   # we don't want that
    return negativeLogLikelihood


def fit(x, shift):
    bounds = [(0, n+1), (1, 700), (1, 700)]
    dist = stats.betabinom
    opt_res = differential_evolution(func_nll, bounds, maxiter=100, tol=0.001, seed=45,
                                        args=(dist, x, shift))

    return opt_res


x = [28, 67, 69, 69, 71, 69, 55, 65, 68, 67, 67, 68, 69, 60, 40]
n = max(x)

mode_value = Counter(x).most_common(1)[0][1]
n_zero = 4288  # high number of zeroes
is_zero_inflated = (n_zero >= mode_value)

shift_loc = 0
if is_zero_inflated:
    shift_loc = 1

b_param = fit(x, shift=shift_loc)

print(b_param)

在scipy==1.8.1环境下，代码输出：

df = fun(x) - f0
     fun: 49.59657327021416
 message: 'Optimization terminated successfully.'
    nfev: 2802
     nit: 61
 success: True
       x: array([70.00205357,  6.28931939,  1.0005225 ])

升级至scipy==1.9.0后，代码输出：

df = fun(x) - f0
     fun: inf
 message: 'Maximum number of iterations has been exceeded.'
    nfev: 9129
     nit: 100
 success: False
       x: array([ 45.83055307,   3.17350763, 596.7724981 ])

想了解Scipy 1.9.x中Beta-Binomial分布的实现是否存在破坏性变更，导致该问题出现。

内容的提问来源于stack exchange，提问作者2m4t