This data is derived mainly from mostly un-reliable large scale observational/epidemiological studies where confounding variables are common, inaccurate survey reporting occurs, and you get relationships for a bunch of different variables. Correlation DOES NOT imply causation and observational studies can only kind of prove relationships but NOT causal ones.
A plethora of RCTs are needed to strengthen the data from these studies before they can be taken at face value. You have to realize that these studies produce a shit ton of correlations, some more ridiculous sounding than others. They have to be verified as causative relationships, as in real life, there is an interplay between 1000s of variables. Note that they are also combining data from multiple studies with varying methods.
Some confounding variables can be removed, but often they aren't properly removed or the researcher wants to outline a specific correlation and pretend it's a causative relationship to stroke their scientist ego. People should check the variables they took into account. It could just be the fact that meat is energy dense and correlates to weight gain in the population as a whole on a mixed diet.
This weight gain would be what is the actual direct causative factor for the increased risk, or even something else. It could be that vegetarians are at healthier bodyweights or are more health-conscience than your average unhealthy American who eats meat (which is most mostly true in general).
Some other questions that need to be asked:
Is there a real dose/response relationship between meat intake and diabetes with all other factors controlled? ---> You need lots of RCTs for this.
Is there a strong theoretical basis for meat directly contributing to diabetes when all variables are taken into account?
Does meat contribute to health and disease in a way that is heavily dependant on the individual's context (this is what is usually the reality)?
Also, even if one is to accept the lack of strength this data has at establishing causation, the best relationship they were able to make was a 10% increase in diabetes risk with meat consumption, which is notable but not huge.