I think the way Google were blocking it before was the the httpWebRequest that the Manifold google ISI driver was generating didn't have any client information included in it. This made it very easy for Google servers to refuse it. All I did was to add the same client information my browser supplied.
There are a few other bits and pieces that the browser supplies to google when getting the images, which I could probably also add into the request. As it is, Google could easily figure out that it isn't a standard web browser request, but it depends on what sort of stuff they have set up on the backend to detect this type of use....
If it does boot you out after extended use, let me know, I can have a go at adding more stuff into the request that may help.