… I’m a plane guy :o 
Bumped.
Are there any confirmed incidents where a submerged sub observed nearby enemy warships in WWI or WWII communicating via signal lamp, semaphore etc. and decoded the message(s)?
I don’t know of any instances where a sub had a mission to do that. And why would they even bother? Flashing lights would help locate the ship, and after that they would sink it or maybe reposition to a better attack position.
Also, I’d think there would be nothing worth decoding sent by blinker. Anything so important would be encrypted and radioed.
Classic periscopes aren’t actually all that long. You have to propagate the image through several “generations” of relay optics, and even with anti-reflection coatings you inevitably get some loss at each optical interface. Periscopes were effectively limited to several tens of feet in length, well under 100 feet.
Nowadays, you can basically put a Camera on a Stick instead of using an optical relay system to get the image down to the interior of the sub, and the length is limited by how much of that stick you can store in your sail (and still stand up to running at sea). They call it a “Photonics Mast”, now. I don’t know how long they can be, but I’ll bet they make them longer than the older periscopes used to be.
About maintaining a constant depth: The density of water changes only very slightly with depth, essentially not at all over a span of mere tens of meters. So if you’re at neutral buoyancy for periscope depth, you’ll also be at neutral buoyancy for a depth well above or below that, and it’d take very little to move you upwards or downwards. With quick enough responses on the ballast control, you could in principle dynamically alternate between being positive and negative buoyancy, to oscillate around your target depth, but I’d be surprised if they can respond quickly enough to make that practical. So in practice, it’ll usually be easier to either run slightly light and use downward “lift” from the control surfaces, or to run slightly heavy and use upward lift, to maintain position.
You’re correct that the density of water is a small effect. As you go deep, however, a much larger effect is the decreased buoyancy due to the slight compression of the submarine hull itself (which is a large effect because the hull is so large). As the mnemonic goes, “the deeper you go, the heavier you get.”
At periscope depth, on the other hand, the largest factor by far is the Bernoulli effect as the submarine approaches the surface. As a moving sub approaches the surface, the fluid velocity directly above the sub tends to increase, resulting in a lowering in the fluid pressure in the region above the sub.* This results in the sub being pulled or “sucked up” to the surface. The effect is magnified as the sub’s speed increases, and is also magnified with rougher sea conditions (higher sea state). To avoid broaching (unintentionally surfacing), subs have to take on increased weight in their variable ballast tanks prior to going to periscope depth. They then maintain depth by using the control planes (which obviously only work due to the forward motion of the submarine).
Under normal operations, it is indeed not generally practicable to manually maintain depth control at periscope depth solely by using variable ballast tanks. That said, ballistic missile submarines have a need to compensate for the large change in weight when launching missiles (and the subsequent flooding of missile tubes), and so use a separate system of automatic depth control tanks.
*This same effect tends to pull surface ships on parallel courses in close proximity towards each other, which is an issue (for example) during underway replenishment operations. Again, the effect is magnified as the ships get closer together, and as speed increases.
And that would be an example of the “very little” that it would take to pull you out of your chosen depth.
Actually they use “extremely low frequency” (ELF) signals to communicate and I think it is only one way (the submarine cannot talk back using ELF). IIRC the transmitting antenna for this is in northern Wisconsin and uses 25+ miles of antenna spread over several acres. I also think it is a lot more power hungry than one might guess to run it. And yeah, it is very slow to communicate this way. Only really short messages are sent.