For new use, NiCads have pretty much vanished from the market, but properly looked after, good ones have a heck of a long life expectancy. The big problem with storing them was that they would self discharge, and if they did that rather than be discharged, different chemistry happened, and they wouldn't recharge properly. The correct method of storage was to discharge them fully, then before use - WELL before use - to bring them back up to full charge by reforming them by repeated charge/discharge cycling. Not something that the average modeler would be inclined to do.
I suspect that the alleged "memory" was nothing more than inconsistent manufacture of the individual cells, and that if you combined poor storage with one weak cell in a pack, you soon got to the stage where one cell would not charge properly, and you had a weak link in the chain. As soon as the incompletely charged cell became discharged under load, the current being drawn from the pack would try to reverse charge it, dropping the pack voltage by two cells-worth suddenly.