The pool does this by using unique extranonce/coinbase data for every connection. It also checks for duplicate submissions.
What about the S1 itself, or any other large mining equipment? Has anyone reputable audited the output of the S1 to make sure all is copacetic? 200gh/s (OCd) is a lot of hashes to verify everything is working right. Has anyone fed it controlled data and verified the output is sufficiently random to indicate it's working properly? I think I know enough about mining works to think that a hash is a hash is a hash, so if it appears to be generating good hashes, presumably at sometime it would generate a hash that equates to a block?
A hash is a hash is a hash. In fact, each difficulty level 1 mining pool share is equivalent to finding a block at difficulty level 1. It's trivially easy to determine that any given share is valid, and in fact, all mining pools do this for all shares. Shares that are invalid are rejected and don't count towards your indicated hash rate. So your indicated hash rate is your real hash rate.
There's nothing to worry about.