Fit does not throw exception if negative event times are passed

[sebp]

CoxPHSurvivalAnalysis can be fit with negative observe time points. It only fails during prediction. However, by definition time should be non-negative and fit should throw an exception.

import numpy as np
from sksurv.linear_model.coxph import CoxPHSurvivalAnalysis
from sksurv.util import Surv

X = np.array([[1, 2, 3, 4, 5, 6, 7, 8]]).T
y = Surv.from_arrays(
    event=[True, False, False, True, True, True, True, False],
    time=[0, 1, 2, 3, 4, -1, 6, 7],
)

# successful
cph = CoxPHSurvivalAnalysis().fit(X, y)

# fails
cph.predict_survival_function(X[:2], return_array=True)

Traceback (most recent call last):
  File "/home/sebp/Projekte/scikit-survival/cph_neg.py", line 15, in <module>
    cph.predict_survival_function(X[:2], return_array=True)
  File "/home/sebp/Projekte/scikit-survival/sksurv/linear_model/coxph.py", line 618, in predict_survival_function
    return self._predict_survival_function(self._baseline_model, self.predict(X), return_array)
  File "/home/sebp/Projekte/scikit-survival/sksurv/base.py", line 49, in _predict_survival_function
    return self._predict_function("get_survival_function", baseline_model, prediction, return_array)
  File "/home/sebp/Projekte/scikit-survival/sksurv/base.py", line 26, in _predict_function
    arr[i, :] = fn(times)
  File "/home/sebp/Projekte/scikit-survival/sksurv/functions.py", line 95, in __call__
    raise ValueError(f"x must be within [{self.domain[0]:f}; {self.domain[1]:f}]")
ValueError: x must be within [0.000000; 7.000000]

Discussed in #404