MAINT: minimize_ipopt: fix late binding bug when defining constraint Jacobians

[mdhaber]

As reported in #194 (comment), minimize_ipopt fails on the scipy.optimize.minimize example with three constraints. This is due to a late binding gotcha when the constraint Jacobians are defined in a loop. This fixes that bug and demonstrates that the example problem is solved successfully after the fix.

Try me before and after.

from cyipopt import minimize_ipopt

fun = lambda x: (x[0] - 1)**2 + (x[1] - 2.5)**2
cons = ({'type': 'ineq', 'fun': lambda x:  x[0] - 2 * x[1] + 2},
        {'type': 'ineq', 'fun': lambda x: -x[0] - 2 * x[1] + 6},
        {'type': 'ineq', 'fun': lambda x: -x[0] + 2 * x[1] + 2})
bnds = ((0, None), (0, None))

res = minimize_ipopt(fun, (2, 0), bounds=bnds, constraints=cons)

Incidentally, this also fixes an unreported bug: kwargs was being passed to approx_fprime. but approx_fprime doesn't accept kwargs. Another instance of this bug is fixed in https://github.com/mechmotum/cyipopt/pull/200/files#r1198383113.

[moorepants]

LGTM