Reverberation is a surprisingly deep subject. It's everywhere in our lives all the time, although we're often unaware of it because our brains are processing it subconsciously. But take it away and suddenly we're in a strangely disorienting environment. Most people can't stand being in an anechoic chamber for more than a few minutes.
When thinking about artificial reverberation it's good to start by thinking about real acoustical reverberation and what it means to us for depth perception and location, as well as the aesthetics of musical sounds made more pleasing with the addition of time-based complexity. When you insert a reverb plugin, you're leveraging the listener's real-life experiences to create an aural illusion that means something to him, even if it's on an unconscious level.
A violin without reverb is fingernails on a chalkboard, but the violin was designed to incorporate the room as part of the instrument. Concert halls are carefully crafted for maximum diffusion density and spectral uniformity while keeping delay times short enough to not impinge on clarity. When you employ a digital reverb, it needs to partner with the instrument or voice in a similar manner.
OK, enough philosophizing. On a more practical level, and concerning vocal recording in particular, your best bet is to start with an acoustically-neutral environment. Use acoustical absorbers around the microphone and singer to remove as much of the room sound as possible. The result will be unnaturally dry and you probably won't like it, but it will provide a baseline from which to add artificial reverberation for any number of effects.
Part of that is panoramic. When someone sings, even unamplified, their voice is a monophonic point source. However, that's not how we hear it, because we're also processing reflections that come from many angles. That's what you want to duplicate. Except that instead of relying on what reflections the room provides (which is usually crap) you now have complete control over it.
"True stereo" reverbs are irrelevant in this case, even though the end result is a stereophonic signal. That's because the source is monophonic and only the reflections are arrayed across the panorama. For this, PerfectSpace does a fine job. As it generates "reflections", it's panning each of them differently, imitating the effect of natural reverberation in a room. The main signal remains in place, but the perception is that it's coming from somewhere in a real physical space.
"True stereo" reverbs actually employ separate left and right reverb engines. They can enhance the spatial illusion when their input consists of multiple voices or instruments panned apart. A sound source at the far left of the room would not deliver the exact same reflections as another source at the far right side. The left source will favor early (short) reflections from the left, while its reflections coming from the right will be longer and quieter.
For that scenario, I would only bother with a convolution reverb if there was a particular IR that I wanted to use. There are some unusual IRs available that produce effects that are not possible with an algorithmic reverb. PerfectSpace in particular has some cool features that are rare on algorithmic reverbs. But for something mundane such as a vocal bus, algorithmic reverbs do just as well and are much less CPU hungry. Some give the option of mono-in-stereo-out or stereo throughout. Most are capable of both natural ambiance and surreal exaggeration.