-Pressurized water takes 1 tick to move 1 square.
May or may not be true. How is the water "pressurized?"
With a pump? The pump would have to be hooked to a constant 7/7 water source (like the bottom of a very large cistern). The pump would put 7/7 water into it's output tile, but 1/7 can still flow off from that in the same step. It's not a big problem for a 2-tile long repeater, but after adding a certain length, you can't ensure that 1 tick/1 tile holds true anymore; it could move slightly faster (7/7 sliding water riding on top of 6/7). A large cistern would be more reliable as it teleports water in a single step, so long as you have enough surface tiles in the cistern, but it becomes infeasible after a certain length.
But, that means nothing, as the second stage of the repeater uses depressurized water. As in, the water flows randomly and may take any number of steps to flow off the plate and open the bridge. And, of course, there will already be some water in the repeater when the bridge is next opened, which may complicate things even more.
> >
P|+__...__ (+^|)
> >
What's the purpose of the drains? If that draining corridor is 38 tiles long, it adds much more than 38 ticks. Consider the drains are closed (via hatches). Water needs to fill 38*3, or 114 tiles before hitting the repeater. Yes, water can reach the repeater before the corridor is full, but when you're going for a certain depth, the random flow is much more pronounced given a larger space, and it will kill any determinism.
Another design might be
P|+ + ... + (+^|)
where those drawbridges in the middle are lowered while the corridor is filling, so you only have 38 tiles to fill, then smash all that water when the bridges raise, to clear the space for the next cycle. That doesn't overcome the flow limitations, but it will be more reliable, unless I made a wrong assumption about the drains.