That would be impossible as the LOAD as given in linux/unix has nothing to do with the cpu load of a server but the total of processes queued (waiting for their turn). If their main node has such a high queue count it should be fairly easy to see for them which process is causing that. No idea why some hosts keep sending their customers to fairytale land every time when it comes to linux load information. Anyways; Exim with that many processes doesnt sound to good either.

As for one VPS bringing down the main node; this is really not that strange as you keep in concideration that a VPS is still a node with many shared customers. Where some things can be regulated, like cpu cycles and memory, but if your node goes completely balistic on its disk IO it will bring a VPS-main node to its knees.