Suppose a pack of 8 AA NiMH cells are serially attached and its known NiMH cells should not go below 900mV per cell.
Does this really mean the pack should not be discharged past 7.2v?
Your sums are right, 900 mV * 8 = 7.2v.
However, what happens if the pack is unbalanced, has 7 cells at 1.0 v, and one cell at 200 mV? That measures 7.2 V at the pack terminals as well.
The cells may not start out this unbalanced, and of course they get charge balanced on each full charge. However, as they wear out, they may wear at different rates, and their capacities may drift apart a little.
Although the 200 mV cell is completely exhausted, it's likely not yet permanently damaged. That will happen if you use the pack for just a moment longer, and the cell goes down through 0 V, and keeps on going, getting 'charged' in reverse by the other cells. Note that's applicable to nickel chemistries only, lead and lithium have different damage voltages. Once you've reverse biassed a cell, the battery pack becomes more or less unusable.
When you're using a battery of series cells, without access to the internal cell voltages for monitoring, you need to have some extra cushion voltage to allow for cell mismatch. The endpoint voltage you choose will result in a tradeoff. A higher endpoint results in less usable capacity per cycle, but more expected cycles lifetime before you damage the pack through this mechanism.