So the question is this - why does the NH pool request work restarts so often? Even when I'm hashing the same order (I know for sure because there was only one active one and all miners on the pool were assigned to it), the pool requests work restart every 10 seconds or so. This has dramatic effect on the average hash rate. For example I have a really stable 3.5 TH on regular pools, but on NH with all the work restart requests, it drops down to 2.9-3.1. So this actually means that any increase in pay-per-GH over a regular pool will be negated by the significant decrease in your rig speed.
Couldn't it be adjusted so that there is no work restarts unless you change the order you are hashing for, or something similar? See the screenshots, work restarts 10-12 seconds apart.

What you are seeing on your miner is actually what the buyer's target pool is sending to you. Our system does not produce work restarts since we are forwarding work from buyers target pool to your miner. Some buyers are targeting orders on some pools which are generating lots of work restarts. This is usually a symptom of multipools which are switching coins very fast and don't have an efficient implementation of coin switching. In some cases this is also a result of some custom private single-coin pools. We addressed this issue a long time ago so that we reward miners (sellers) with extra shares when fast work restarts happens. So it is
very important for you to monitor the hashrate, reported on our website (it is calculated from shares, rewarded to your miner) and the actual earnings, not only the hashrate that is displayed by your miner, since in this cases you will be rewarded for more then your miner is showing you.