It seems that pages under the root get's it's siteID set to the default site's id during indexing. Is this by design or a bug?
Randompage will have it's siteID set to site1 in the index.
Can't talk for EPiServer but I would guess it is by design with the new strict routing that the site with "*" in hostname should get all pages outside a start page.
Now there must aways be a reverese route to an absolute url for a page otherwise you get really strange exceptions. I know you can get really strange exceptions if you forget to have a "*" setup in site settings anyway.