Have you guys actually tested this? You're talking about it like there's no scaling factor.. but there is!
.01-9.99 manufacturing will give you ONE of the listed bonus. 10.00-19.99 will give you TWO of the bonus. Just load up the game and try it.
As far as research and economic projects go, you might as well not have a scaling factor involved as, unless you're on a manufacturing-specialized world, setting the sliders to a position such that you get more than the minimum project bonus will hurt you more than it'll help you. The net change in the output multiplier from using a project relative to setting 100% of the planet's production towards the desired output type is
[net change] = (1 - [manufacturing fraction]) * 0.1 * ceiling([manufacturing fraction] * [production] * (1 + [manufacturing bonus]) / 10) - [manufacturing fraction] * (1 + [base output bonus])
where the base output bonus is the total bonus to the output type corresponding to the project from all sources other than the project, and ceiling(X) is the function which returns the smallest integer N such that N >= X. An alternative way of looking at this is that for a given project bonus P and a given manufacturing fraction x, the maximum bonus B to the output type corresponding to the project type from non-project sources you can have before using the project becomes harmful is
B = (P / x) - (1 + P)
Much better to have used the traditional "converts manufacturing into research or economy" style projects.
The projects used to do this, but the way it was implemented resulted in the output being multiplied first by the manufacturing multiplier and then by the (research or economic) multiplier, and so people complained, and so they changed things to the present system.
it's 10 raw research multiplied by 40% (res building) + 46% (46 mfg converted by research project) - 18.6 research.
This is actually going to give you 19 research; the 46 manufacturing gets turned into a 50% bonus because it's run through a ceiling function after being divided by 10 but before converting to a 10% per X bonus. You actually maximize at 1% manufacturing in your example, with 29.7 research produced; somewhere between 6% and 7% manufacturing, the total output falls below 28 and the project begins hurting you.