1. You could do that (in theory), but you would have to make sure, all of these 1000 directories are strictly of the same kind. If some great webmaster renamed some of the IDs of the DOM tree, ZP wouldn't know where that button or field is, and thus can't create the account there. But that's how it works. You can do some tricks to look for the fitting button (watch for name, # in hierachy, DOM text, source of page, rendered text (what you see)), but afterall, a smart webmaster could rename all the stuff, so you can't access the page with a 1000-sites bot. You would have to do a single one only for him

But thankfully most webmasters are a bit lazy (or frightened of screwing the installation), so they use standard templates, which you can access easily. Yestrerday I started scraping for german bookmarking sites, I let ZP look for a certain feature to make sure, they are all of the same kind. And a first test was very good, no matter how the site looked, ZP was able to go through the setup. That makes nearly 90 bookmarking pages at my fingertip.
2. Captchas work very good. For easy captchas (not reCaptcha) you can train a module which will solve the captchas for free.
3. The bot does, what you tell him to do. But you have to learn it...
Regarding the article: what tells you, if the article was accepted? If it's an email, let ZP look for the email and parse it for some words like "Congratulations, your article has been accepted!", if it's inside your profile, do a template that goes through all your article directory profiles once a day, and tells you, which articles are posted (let it write to a file).
You can't sell bots with ZP. It's like having help with your work. It's not about selling your methods. You could sell services, e.g. I could offer a German bookmarking service with that new template. Or you could sell twitter, email or whatever accounts.
I guess with some (hard) work, you could build most of these 57$ or 97% or 147$ tools yourself (with a much smaller footprint).