My understanding about the "free memory" part is not that the analyst should not remember anything what the patient has told him but rather that what has been said previously is not going to determine anything in the future sessions. Every session starts a fresh with new opportunities.

My own analyst draws a lot from Bion and neo-Kleinians.
Sure, he analyst remembers many things I've told him but these things don't define me for him and he does not base his expectancies on those things. In that sense every session starts afresh.