It's not uncommon for different parts of a system to be powered by different supplies which share a common ground. This may be because some parts need 3.3 volts while others need 2.0 or 5.0, because some parts may need to be powered on and off separately from others, because some parts may generate a level of electrical noise on their supplies which other parts would be unable to tolerate, etc. In some cases, the circuitry which generates a reset may not operate or be controlled by the same supply that operates the CPU. Having the reset generator on a different supply from the CPU is not a problem if one is using an active-low reset and either the CPU can tolerate voltage levels above VDD or the reset line can be weakly pulled high by something attached to the CPU supply.
As a simple example, imagine a 3-volt CPU which is interfaced with 5-volt chips. The external circuitry will malfunction in arbitrary fashion if VDD drops below 4.75 volts and would require reinitialization after voltage rises above that point. The CPU itself might be able to run code just fine if the main supply voltage drops to 3 volts, but might not be able to do anything useful; the cleanest way to ensure that the external hardware will get initialized after VDD rises above 4.75 volts will be to reset the CPU whenever VDD is below that point. Using an open-collector reset chip and a passive pullup to the CPU's VDD would be the simplest approach.
About the only disadvantage to that approach of handling reset is that a passive pull-up will consume current continuously while the system is in reset. In systems powered by mains, energy storage devices [capacitors] expect to be drained completely dry without damage. In systems powered by rechargeable batteries, however, draining current from a discharged cell may cause excessive wear. Even in systems powered by disposable batteries, continuous current draw may undesirably increase the risk of batteries "venting" [spewing goo].