Everything below is based on columns as output and rows as input. If you do the opposite, revcerse the staements.

to detect simultaneous press you need a series diode for EVERY switch
True, but if you only need one (or a few) you can detect simultaneous in same row. I use this technique to have a "secret code" or two for field engineering purposes.

I just remembered one thing about keyboard scan without diodes: If you do this DO NOT assign the outputs as push-pull (on devices where that is possible), some (accidental) simultaneous presses can blow the ports in that case by connecting two column outputs. I routinely assign pure outputs as push-pull, but saw the problem in this case before the circuit was powered.

Erik