I'm sorry to say, this sounds like Keyster needles in the slides. This is precisely their symptom, because they taper too much, which lets the engine run very rich in the 1/4 to 1/2 throttle range. They are also often a little too long, and too thin. This makes them run very rich above 1/2 throttle, so the engine drowns in 10:1 rich mixtures when the slides are pulled wide open: you would need 25% more airflow and 60,000 volt coils to fire that (think turbo, CDI, Mallory coils...).
Pop: first thing to try - remove the carbs, pull out a needle, verify if the number on it is "27201" or "2701" (with possibly an "A" or "B" behind it). The former is Honda's, the latter is Keyster's. If Keysters, and you wish to ride again before you find some genuine Honda needles, drop the needles 2 notches in the slide (move the clip toward the blunt end 2 notches) and set the idle screws at about 7/8 to 15/16 of a turn. The CB750 carbs go leaner as the screw is turned in, richer as it comes out. This will leave you with a pronounced flat spot at about 65 MPH, but will at least keep you from running out of sparkplugs.
I have used common 1/2" wide hose claps on hard carb boots for many years, with good results. They are not as pretty, but do a good job of removing vacuum leaks. The wintergreen and xylol treatment I have been doing to (over a dozen of) them now works perfectly, making them like new! But, the old clamps do stretch. You can sometimes add some grip to them by installing a small washer on the screw (might have to nip off the sides of the washer to make it oval-ish to fit): I have done this many times, too.
Aaarrrggghhhhhh. Hondaman, how I wished you were right about the needles, but no, all four are marked 27201. But hey, I'm getting better are carb rack removal; 22 minutes this time.
So just a little more data. The slides themselves are marked (stylized K) 25 and also 102 on the bottom of the slide where the needle protrudes. The 105 main jets I first installed after rebuild are marked with the same stylized K just before the 105 mark on the sides of the jet. The three 95s that came with the bike are marked on the top of the jet with RO along with the 95. There's lots of "wiggle room" with the needle in the needle jet when the slide is near fully up, less when half down, and just barely moves in the needle jet when the slide is open just enough to get to the needle to test. I checked the float heights again, all the same, 26mm when measured from higher point on the side of the upper bowl with the tang just kissing the shaft of the needle, and 25mm when measured from the lower slot in the side of the upper carb bowl. The slow jets are tight, the main jet holder is tight. When synced, the readings were in the 22, 23 range on Carbtune device.
Once again, needle clip is in second position from bottom. The plugs I was running before the carb clean with 95 jets were NKG D7EA and did not foul nearly as much as the Denso X24ES-U with either the 105 or 95 jets. And all things being equal, why would No. 3 plug look normal when the rest were badly carbon fouled?
Clearly need help, what next?
If I might ask a seemingly unrelated question:
What is the compression of all 4 cylinders? A tight intake valve on #1 will make #3 show up rich, interestingly enough, while #1 will show leaner by comparison (and, this is the same all through the firing order, BTW).
I'm sure of one thing: #3 will run darker with the retarded timing you mention. #2 will eventually also show darker, but #3 does it first from late spark. It sounds like you might have some bad (geomertically speaking) points, or points plate (can it move around more than 0.004" between the "posts" when the 3 screws are loosened? That's too much.).
But, late spark usually takes about 50 miles or more to foul plugs.
Vacuum leaks at the carb side of the hoses will make the carbs run in the richer idle range, all the way up to 1/2 throttle or even more. This is because on the carb where the leak is, the air is moving slower than it should be under the slide, since some of it is coming in downstream of the carb. These carbs run about 12:1 mix at idle, which becomes about 14:1 at 1/4 throttle when the cutaway on the carb slide becomes active: if the vacuum leak is, say, 10% of the airflow, then the idle circuit is still working in the affected carb until almost 1/3 throttle. When this happens, the other carbs start leaning out like they should, but the leaky-hosed ones are still running "slow" and richer. Anything richer than 14:1 will darken plugs quick: the only reason it doesn't happen so fast at idle is because very little air is actually entering the engine, and the burn time is very long.
Vacuum leaks on the head side of the hoses will usually cause erratic idle or cold-blodded tendencies. Here, they don't affect the color of the plugs so much. They just make the engines run unevenly, and generate "flat spots", one of the hallmarks of the Mid Fours and the Baby Fours (they have O-rings here which often leak).
The float levels sound OK: they run well all the way up to 24mm (my "racing" setting), if leaky there a little around the bowl seals on the sidestand.
I'm placing my bet on these things, since it sounds like the needles are right, and if I were there, would do them in this order:
1. If the condensors are more than 3 years old, change them. They leak after a while and sap spark.
2. Clean and polish (especially those #2) points, regap and retime to get everything spot-on at 0.014" gap and "F" marks at idle speeds under a timing light. Make sure the spark advancer goes to full advance by 2500 RPM for both sides.
3. Ohms-check the sparkplug caps. If they are more than 500 ohms different between the caps on one pair (i.e., the 1-4 pair or the 2-3 pair), or if the resistance on that model is more than 7.5k (7500) ohms, replace them, and in pairs. Use 5k ohm NGK caps. Make sure they screw tightly into the spark wire: if not, cut off a little bit of the wire(s) (like someone said above) and screw them in tighter.
The above is essentially an electrical tune-up to make sure your spark is right. It's real hard to measure it directly...
4. Get 1/2" wide hose clamps and install them on both ends of the carb hoses, tightly. If this fixes the hoses, they may need more wintergreen, or maybe just replacement. I recently saw 2 cracked ones on a local 750 at a shop: both those cylinders were fouling plugs (per my explanation above). They were cracked through to the inside. almost the length of the bottom of the hoses.