The receive gate timing being generated is based on gates 62 and 58. McEwan's purpose isn't to have no jitter in the xmit pulses, but to reduce 0.1-10Hz drift in the T-R gate delays. Since 58 and 62 are in the same chip, it doesn't seem that unreasonable - except for the title which says "precision". His ability to detect the radar reflectivity changes within a 2m radius only needs a 13ns delay, and jitter of +-3ns isn't going to be noticed, unless there's much variation in the average T-R delay over 0.01-1 second intervals.
Not worth patenting perhaps, but functional for his purposes...