[local state-averaging] H1EZipFCISolver member functions

[MatthewRHermes]

To enable local state-averaging, the H1EZipFCISolver member functions must be further generalized. Currently, in the dev branch, the *_make_rdm* functions will do local state-averaging correctly thanks to this hacky function:

mrh/my_pyscf/mcscf/addons.py

Lines 63 to 84 in ef4bf4a

	# This is a hack designed to, i.e., set
	# self.fcisolvers[i].nroots = 1 for all i
	# in the context of StateAverageMixFCISolver._loop_civecs, so that local state-averaging
	# in the ith Hilbert space can be totally abstracted away from the HIEZipFCISolver,
	# StateAverageNMixFCISolver layer
	def _loop_civecs (self, *args, **kwargs):
	_solver_args = self._solver_args
	_state_args = self._state_args
	for i, solver in enumerate (self.fcisolvers):
	my_args = []
	for arg in args:
	if isinstance (arg, (_state_args, _solver_args)):
	my_args.append (arg[i])
	else:
	my_args.append (arg)
	my_kwargs = {}
	for key, item in kwargs.items ():
	if isinstance (arg, (_state_args, _solver_args)):
	my_kwargs[key] = item[i]
	else:
	my_kwargs[key] = item
	yield solver, my_args, my_kwargs

But kernel, approx_kernel, and the other states_ functions, which use _loop_solver rather than _loop_civecs, generally do not work with local state-averaging.