Main loop phase implementations.
Target: <1ms per iteration Maximum: 5ms per iteration
NETWORK_IMPL_GUIDE.md §6.2, §6.3