SCF workchain should update options when _handle_not_enough_memory took place

[Tseplyaev]

In a situation when options of a FleurCalculation were updated in _handle_not_enough_memory, SCF workchain will submit next FleurCalculation using the old option which most probably will cause the same problem.
Thus SCF workchain should somehow understand that nested FleurBaseWorkChain updated options and use them for the subsequent iteration.