avoid RuntimeError due to nans in overschrijding

[veenstrajelmer]

The function calc.calc_overschrijding() sometimes warns "RuntimeWarning: invalid value encountered in scalar power return -p_val_gt_threshold * (alpha * x**(alpha - 1)) * (sigma**(-alpha)) * np.exp(-((x/sigma)**alpha) + ((threshold/sigma)**alpha))". This for instance happens in test_calc_overschrijding. This happens due to nan values from powers like here. The source is optimize.minimize() and further down the line (sigma ** (-alpha)) (and others) in der_pfunc(). This happens in one of the iterations of the Nelder-Mead optimization function, might be resolvable by supplying additional settings.

Can be reproduced by checking np.isnan(sigma ** (-alpha)) in der_pfunc() when executing:

import os
import hatyan
import kenmerkendewaarden as kw

dir_base = r'p:\11210325-005-kenmerkende-waarden\work'
dir_meas = os.path.join(dir_base,'measurements_wl_18700101_20240101')

current_station = "VLISSGN"
df_ext_all = kw.read_measurements(dir_output=dir_meas, station=current_station, extremes=True)
df_ext_all_12 = hatyan.calc_HWLW12345to12(df_ext_all)

dist_exc = kw.calc_overschrijding(df_ext=df_ext_all_12, rule_type=None, rule_value=None, 
                                  clip_physical_break=True, 
                                  interp_freqs=[1])