Spent the weekend tearing up my roof deck and finally found the leaks

I actually went through something very much like this last year. Seeing the screws go right through the membrane is such a common failure point, even when you think you are careful. When I pulled up the boards, I found that the factory screws were too long for my liking, and they were definatly puncturing the underlying material. My approach for the penetration issue was a bit diffrent than yours. Instead of patching the holes with fabric and liquid, I used a specific polymer sealant designed for roofing around each head. I took the screws out and carefully applied a bead of it where the metal met the waterproof layer before putting the screws back in, making sure they were not overtightened.

The slope was my biggest headache. I found that trying to adjust the deck level after the membrane was laid was impossible. So, I decided to raise the subframe slightly. I cut sections of rigid foam insulation, the kind that is like a thick slab, and placed them under the main support beams on the high side. This created the necessary fall for the water to run toward the drain we had near the corner. It felt a bit unorthodox to stick insulation under the structure, but it gave me the proper pitch, and the water now flows cleanly away. It was certainly a laborious project, but it seems to have solved the issue for now.

I dealt with the same membrane puncturing when I did my patio last spring. It was more of a thin polymer sheet than the WPC boards you described, but the failure mode was identical. I initially just drove the fasteners in, but after the first heavy downpour, I saw tiny beads of water right around the screw heads. It looked terrible and wasn't drying out. Instead of patching every single spot like you did, I found a better way to handle the stress point. I used a heavy-duty sealant under the heads of the screws before securing them. It was a bit unusual, but becasue the sealant expands and adheres to the membrane, it sealed the hole where the screw passed through. This stopped the slow leaks imediately. It felt like a bit of a hack, but it was much less labor intensive than building up the pitch with sleepers first. This prevented any issues with water pooling in those little valleys. It seems the key is managing the force at the contact point, rather than just filling the resulting hole.

The sleeper route is the right call but one thing to think about before you close it up: how are those sleepers getting held down? On my own balcony rebuild I ran 2x4 sleepers on 16 inch centers and glued them down with membrane-compatible adhesive instead of any fasteners. Even one errant screw or nail into the sleepers and you are back to square one with punctures. Also worth a check with a level on each sleeper run before the boards go on, since a slight twist in a sleeper will create its own little pond spot that defeats the whole slope you are building in.

Nice that you did the bucket test before closing up, wish I'd done that on mine. I redid a small flat section a couple summers ago and skipped the water test, then found a pond the size of a dinner plate in the exact spot I didn't check. This week's rain will show you everything, but bring a chalk pencil out there after it dries and trace the low spots. Way easier to pop a board or two and shim a sleeper when you know exactly where it's holding water.

Update: good point about holding the sleepers down without piercing anything, that got me thinking. Mine are just sitting loose right now and I figured gravity plus the board weight would do the job. Now I'm wondering if I should glue them down like that too, or if loose is fine for a small deck. Also grabbed the chalk for the rain test, that dinner plate pond story sold me.