When I tested pyramid-like interspersed blocks, I found that in no way did lava get duplicated by falling onto blocks and splitting apart; it was only duplicated by buckets. X blocks of lava on the input, dropped into a contraption which split and reconcentrated the lava repeatedly, resulted in X blocks of lava at the output.
The thread that was linked to earlier had a design similar to mine which was claimed to duplicate lava, but which didn't actually do so for almost everyone who tried it and reported their results.
So, what I am wondering is this: How exactly is your design supposed to duplicate water or lava? What principle does it operate under? How thoroughly have you tested it? Does it work in both singleplayer and multiplayer? Have other people attempted to replicate your results?
Try to replicate it and see for yourself. I've been doing some "proper" research on this, and have, too, found the "pyramid" splitter ineffective, because you can't split the liquid indefinitely. It eventually just fails to flow or disappears, as my larger experiments showed. What's needed is a way to sustain multiple streams, strengthen them as they split. The single-tile "staircase" has proved to be able to do that.
With the setup on the Magmabomber, one full bucket of input lava gets me about 20 tiles of lava total on the output (i.e. 20 buckets). This is not a direct multiplication by 20 - two consecutive buckets yield ~25 tiles. It's dependent on the duration of the flow, not its strength. And since the bucket-duplication method results in a long series of small half-bucket drops, this multiplier is perfect. When one operation per second is sustained on the input (i.e. two clicks per second, one to fill, one to empty the bucket), the rate of output tops out at roughly 1 tile per
game tick, as seen by the lava level in the little wooden "beaker" with a one-tile hole staying at roughly 1 full tile while lava constantly flows out.
As for other questions, it hasn't been tested in Multiplayer (I foresee a lagfest), and you may feel free to replicate the design and try it for yourself. No need to go for such a large scale, even a small replicator with a four-lane output will give results, I'd expect.