...After a full fill, when the tanks are immersed, we have on occasion observed a tiny stream of bubbles coming from between the tank and the valve. Upon closer observation and removal of the valve we are seeing substantial salt deposits on the threads and 0-ring...
Sounds like the valves are not getting torqued down to specs, and salt water is attacking the o-ring. See the cylinder manufacturer specs for proper torquing of the valve. Most manufacturers are recommending 50 foot-pounds.
...and even salt inside the cylinders....
There are only a limited number of ways that salt water can get into a cylinder: (1) wet fills, (2) breathing a cylinder to zero and hitting the regulator purge valve, or (3) water leaks through the neck o-ring. Which one applies to you?
It sounds like you are getting leaks through the o-ring and cylinder neck. Generally this is caused by improper installation of the valve (i.e., not torquing to specs) or re-using cylinder neck o-rings, or both.
...Worse, we sometimes are unable to remove the valve without very considerable force and in the process have sometimes destroyed both valve and cylinder...
This is a known problem of bimetal corrosion in a salt water environment, secondary to having salt water inside of the cylinder. It's called galvanic corrosion. The University of Rhode Island did a study into this problem back in 1975:
Corrosion of steel and aluminum scuba tanks / Francis C. Cichy, Hilbert Schenck, John J. McAniff, URI Marine Technical Report, 1975
Basically, they found that using a lubricant on the valve threads helps prevents the flow of electrons between the two dissimilar metals (aluminum cylinder and brass valve) and reduces the incidence of galvanic corrosion and sticky valves. The type of lubricant doesn't seem to matter, as long as it is a lubricant that is approved for the intended use (i.e., use an oxygen-compatible lubricant in Nitrox clyinders).
Of course, if you eliminate the electrolye solution (salt water), then you also eliminate galvanic corrosion altogether.
Thread lubricants may not required by the manufacturer, but lubricants do reduce the incidence and extent of galvanic corrosion.
...While the preventive action is to frequently remove and clean the valve/tank and replace the valve o-ring, the various solutions suggested to aid in removal of the sticky valve non-destructively have not always worked to our satisfaction...
That's because you are not servicing your cylinders frequently enough. Based on your statements, you should be servicing your cylinders (i.e., performing visual inspections) at least quarterly or even monthly. If you service your cylinders more frequently, you'll catch the problems before they've developed very far.
...the various solutions suggested to aid in removal of the sticky valve non-destructively have not always worked to our satisfaction...
Once the two metals have bonded together through galvanic currents, your only choice is to use brute force to shear the metals apart.
...Something we haven't tried but which I thought up on my own(!?) was to deliberately introduce about 100ML of fresh water into the cylinder by draining it, turn the cylinder upside down and re-pressurize the the cylinder and see if the water dissolves the salt deposits in the threads from the inside out...
Water isn't going to dissolve metal. Remember, the problem is not "salt deposits," the problem is that you have welded aluminum to brass through galvanic currents. No amount of soaking in fresh water is going to dissolve the weld.
In summary...
I recommend that you:
(a) torque your valves to the specifications recommended by the cylinder manufacturer. I use a long torque wrench and crow foot wrenches:
(b) avoid wet fills and breathing cylinder to zero, and
(c) inspect/service your cylinders at least monthly until you get this problem under control, then consider doing inspections on a quarterly basis.
Valve installation torque values for Luxfer aluminum alloy cylinders and liners - Luxfer: Setting The Standard Worldwide
