The road of electability debates leads to madness.
The GOP wouldn't go easy on Obama, Triumph. They never do.
Hillary's not necessarily more divisive. She generally has high negatives and "would never vote for" numbers, but that's partly a decade and a half of Republican vitriol that Obama hasn't dealt with.
Maybe his numbers would do the same if subjected to that treatment, maybe not. Bill was targeted just like Hillary, but has historically polled better than her (excepting the Monica incident where there was a wave of goodwill towards Hillary, IIRC). But if Obama's the nominee, they will be aggressive, and there are people out there willing to believe bad things about him.
The flipside is that Hillary's willingness to fight dirty -- the reason Frag's backing her -- won't necessarily pay dividends. I think there's a common bias that leads people to assume unpleasant methods of competing are more effective per se.[1]
More specifically, some Democrat activist types give Fairweather Genius Karl Rove and the Swift Boat attacks almost all the credit for Bush's success, and figure Democrats need to copy those tactics to win. Maybe Hillary is more willing to take the gloves off, and maybe that would help. But not everything that's vicious works.
Just like how people tend to overestimate their own knowledge of the candidates' personalities, I think they do the same with their ability to get elected.[2] There's so little data, if any, to figure out how they would fare in a nationwide general election almost a year from now that it just confuses things. Better to vote for who'd do the best job if they made it.
Besides, we all know Edwards was the most electable anyway.
[1] For further reference, see 90% of everything right-wingers write about war, detainment, interrogation, etc.
[2] Yes, there are obvious cases like Kucinich and Paul, but among the front runners it's a lot fuzzier.