... It's true enough that transition times are determined by the technology and not by the clock rate. However.that's a separate issue from the power supply requirement when you're trying to accomplish useful work, as when you have to drive motors or hefty relays, etc. Those require power, and not just 3 volts, though there are some that are quite happy with 3 volts. They do demand current, though.

As for the "solderless breadboards," if you use an older part, e.g. one of the old 12-clockers that has an internal clock of 1/12 the oscillator frequency, the risetimes can also be quite fast, but they settle quickly enough that circuitry designed to operate with 'em won't be too disturbed by a little noise resulting from crosstalk at the pins. The signals are slow enough and persistent enough that external circuitry will settle just as they do. It's not very convenient attaching things to a tqfp or the like, so the signals are often brought out to another device, often just a header, which is not smart if that leads to lengthy wires, but, if it's buffered with a driver of some sort, whether on the Devboard or on the target board, as long as lead lengths are minimized as much as possible, the practice won't cause much trouble.

Conclusion? If you're going to attach things to a Devboard, you should buffer it before going to wires longer than 10 cm, and with DC loading of 10 mA or more. The wires on my 'F120 EVK are long enough at the DIN-41612 connector that I need to buffer the signals before they get to the connector, and, thanks to the connector, probably should be buffered at the target board as well. That adds enough delay that timing might become an issue, so beware! I have a standard I/O arrangement that I use in development, with a CPU board on a backplane into which I can plug other experiments. This would be a problem at 100 MHz, of course, but even at 33 MHz, it's not been a problems so far.

RE