Can you suggest me better literature? Am I using the wrong concepts? Are the references provided above "sufficient"?
I recommend you take a step back and ask yourself what your question is, and perhaps more importantly, why it is you're asking it. Only then will you have a better idea of what you consider sufficient.
I am not satisfied with these references.
Unless I had the actual dataset, then I would find nothing satisfactory. This is because I am a weirdo. Depending upon your aims e.g. consultancy work versus PhD thesis versus simple report, perhaps you feel the same way as I do.
I am an economist working in labour market issues. It is common in my field to divide jobs between those who are intensive in "manual" tasks and those who are intensive in "cognitive" tasks
Nevertheless, if you are an economist, then I suggest your endpoint is not mechanical ability per se, but a far more pragmatic one: the prediction of job performance by cognitive ability. In short, I suspect your actual question can be loosely restated as "what is the predictive validity of cognitive ability and job performance among manual labour employees?"
If I am correct, then you are in luck: this is not a new question. There is good data regarding this matter from the United States Army, partly because they keep very good records, partly because cognitive ability is the basis of eligibility into various military roles.
And you are correct: there is a non-linear relation between IQ and performance; IQ has a high correlation with cognitively demanding jobs (r = 0.80+), while it is far weaker in jobs that are less demanding cognitively demanding (r = ~0.30) e.g. truck driver.
Regardless, IQ remains the best predictor of job performance. Let me dig up some references for you, but I would suggest you don't get distracted by the concept of "other forms of intelligence". In the parlance of biological psychiatry, consider dexterity to be an endophenotype for success in living a successful life as a manual labourer but one that does not fully reflect your endpoint of interest as an economist. This is not to say that I don't believe that they exist, but that as far as you're concerned, perhaps they're thoroughly irrelevant.
EDIT
So, are you actually interested in mechanical ability?
I note that you reference one paper:
Schmidt, F. L., & Hunter, J. (2004). General mental ability in the world of work: occupational attainment and job performance. Journal of Personality and Social Psychology, 86(1), 162.
Having read this paper, I note that it reports IQ scores for various occupations. While I wouldn't say it's precisely what you are after, I would have thought that it's very close in spirit to how I am interpreting your question.
For example, it reports the following correlations:

These occupations are not relevant to your question, but reminiscent of the data that I anticipate you're looking for. So this begs the question: what precisely is your interest? Do you actually care about dexterity, or physical coordination, or do you view them as some meaningful intermediary to your actual interest in work-related outcomes?
Be an innovator
Depending on the type of person you are, and how important the question is to you, you may wish to consider answering the question yourself. To be honest, if I really wanted the data, and the data did not exist, I would not settle for anything less than this.
For example, have you ever considered approaching a particular personnel selection company, or perhaps even better, industry stakeholder to conduct such a study themselves? Here, you would identify a nice, exemplar industry that is of particular interest to you, say, truck drivers, and obtain performance ratings of all the employees. As a pilot study (e.g. n = 20), you would then randomly select two samples, one from the the bottom quintile and one from the top quintile of job performers. Perhaps the bottom quintile would represent workers who have or are about to be fired.
You could then perform a computer adaptive test (CAT) of ability and get a fast and accurate estimate of IQ; if this is worth pursuing further, you would hope to see a robust difference in cognitive ability between the two groups. Adaptive testing applies item-response theory to get a more accurate measure of the latent construct (intelligence) than classic test theory because the questions that are presented to the subject are contingent on their running estimate.
The benefit to the funder of this research is that they would have a financial incentive to predict job performance; as an economist, I am sure you can tell me the difference in productivity by way of $$$ the top performers earn a company relative to the average performers. Just remember, it may be illegal to hire someone based on IQ, but "aptitude testing" is acceptable.
Job performance as a function of baseline cognitive ability
Also, an interesting counter-claim to consider is that while there might be little correlation between cognitive ability and job performance in manual labour employment, there could certainly a floor level of ability required for performance . I must admit that I have not previously entertained such an idea, although it is certainly perfectly tenable:

Cook, M., & Cripps, B. (2005). Psychological assessment in the workplace: A manager's guide. John Wiley & Sons.