When i look at the demo video, i see that the entire section of vocal slips so that thestart point is also effected, even though the start point was correct to begin with.  Is the Vocalsync a visual aid for alignment, or does it elastically stretch and compress parts to match the way V-vocal works?  TIA.

From Sonar's online help: "VocalSync applies varying amounts of time stretching or compression to the dub audio"
 
 
 

Yes, 'Vocal Sync' applies stretching to the track to bring it in line. I've found that the tracks have to be fairly close to begin with or you will get artefacts showing up. I had to do some editing first to get the tracks close, and then 'Vocal Sync' will work well.

Also it's very important to note that the "sweet spot" on the dial will vary for different material. It is NOT a situation where turning it up tightens and turning down loosens. The sweet spot could be anywhere along the dial's continuum.
 
In the "helpful abuse" category, you can also set it "wrong" and spread vocals further apart to differentiate them more if they're too tight.
 

Thanks guys.  Funny though, that it looks to me as if the fist syllable of the phrase in the video gets more out of alignment after the process...
 

Me too, Quadcore.  Suppose you had the following phrase you were trying to double:
 
"Spitting Dan, he was the spittenest man
He'd spit on the floor, he'd spit on a whore, he'd spit on the unsuspecting"
 
Take the word "spitting".  That has a start ("sp") and an end ("ing") but it also has an internal transiet "itt".  In practice, does VocalSync only sync up the beginning "sp" and the end "ing", or does it also try to line up internal transients?  I.e., does it REALLY work, or does it only get you part of the way there.

It's a quick means to get two vocals that are just a little out to line up as best as possible. It's not aimed at perfection.

mudgel
It's a quick means to get two vocals that are just a little out to line up as best as possible. It's not aimed at perfection.

It's also a technology, sort of like Melodyne is a technology, and it can be "taught" to deal with more use cases if Cakewalk wants to continue developing it (which I believe they do).

I guess I'd need to try this out...but the examples all assume you want to tighten up a doubled (unison) line.  Would this work on tightening up timing on vocal harmonies (i.e. dissimilar voices singing dissimilar notes, but at what's supposed to be the same time)?

Noel told me (at NAMM 2015) it is not a pitch tool. It will work best on smaller sections and at present it's best suited for doubled and harmony lines of reasonable duration. 
 
As stated by Craig, this tool will most likely be improved upon over time and will fill a gap (for me) that could end up being a lot more work if I tried to do it in Melodyne.
 
FWIW, in the past when I was faced with harmony BGVs that had a single word that was not aligned I just did a cut of the phrase, nudged it then time stretched. VocalSync could be a time saver even in this scenario.