I have a whole lot of contacts that have clearly been harvested from spam, courtesy Bug 714943 and stuff that still makes it through my inbox that look like this:

> "mike@vee.net" <obvious@spam.com>

I.e. the address label is my email address, but the address itself is spam.

It might be worth examining an existing contacts database to see if there are any legit uses.

If not, before saving a new contact we should try parsing the label and if it looks like a valid email address, only add a new contact if the normalised label is the same as the normalised address.