Why do rocket engines have a nozzle which first contracts to a “throat” and then expands into a big “bell”?
In the diverging section, the walls of the bell reflect molecules of the expanding hot gas(*) in the direction of the outward flow and so increase the speed of that flow. If the flow was sonic at the throat then it will therefore become supersonic (but not necessarily hypersonic) relative to the speed of sound at the throat. Since the pressure drop also leads to a lower speed of sound, the flow may now indeed be hypersonic (usually means Mach #>5 but would have to be at least 2) relative to the speed of sound within it.
*(Note: This only works for a gas and not if your fluid is a liquid, so there is no value on adding a diverging section to the end of your garden hose or pressure washer.)
The accelerating effect of the bell works for any hot exhaust gas at any rate of flow, but for flow rates below the speed of sound in the gas it is less than that of reducing the cross sectional area of the flow, so the bell is never added without first having a converging section to bring the flow rate up to the speed of sound.
For a gas, the effectiveness of reduced cross section is less than for a liquid because the rate of mass transfer can also be increased by compression and it falls dramatically as the flow rate approaches the speed of sound. But I don’t right now have a good explanation for why the speed of sound is the exact cutoff for where constriction becomes less effective than having a bell.