Additional parameters for "scipy.optimize.minimize" in ParametricRegressionFitter is not functioning correctly #1651

Hello, I am currently using Python 3.7.12 with the lifelines module version 0.27.8.

While implementing a custom RegressionFitter using the SLSQP method, I encountered an issue where the bounds parameter in scipy.optimize.minimize does not function properly.

This issue arises because the relevant parameters in lifelines, namely _scipy_fit_options and fit_options, are both passed to the options parameter of the _minimize_slsqp function within scipy.optimize.minimize. Since _minimize_slsqp already accepts bounds as a keyword argument, this results in the error message:
_minimize_slsqp() got multiple values for argument 'bounds'.

To resolve this, the _fit_model function in lifelines should be modified so that the bounds parameter is added explicitly as a keyword argument rather than being included within options.

My modified code to fix this issue is as follows:

options = {**{"disp": show_progress}, **self._scipy_fit_options, **fit_options}
bounds = None if options.get("bounds") == None else options.pop("bounds")
results = minimize(
	# using value_and_grad is much faster (takes advantage of shared computations) than splitting.
	value_and_grad(self._neg_likelihood_with_penalty_function),
	_initial_point,
	method=self._scipy_fit_method,
	jac=True,
	args=(Ts, E, weights, entries, utils.DataframeSlicer(Xs)),
	bounds = bounds,
	options=options,
	callback=self._scipy_fit_callback,
)

Most users may not use this parameter frequently, but an update may be necessary in the future.