200ms is an extremely aggressive pace for any driver based on a TCP/IP query/response protocol. A single packet drop in your network will stall for long enough to invalidate your subscribed tags. Also, you might also need to consolidate your subscribed booleans to make driver optimization possible. For one, do not use alias tags in your processor. Or at least, never access by alias externally. Next, avoid directly accessing elements of AOIs.
You might consider switching protocols, too. I’m a bit biased on this topic, but high speed monitoring like you are attempting is what spurred the creation of my Ethernet/IP module (I/O protocol). It uses UDP traffic to be relatively immune to packet drops, and can reliably run in Ignition in the 100ms range.