The idea makes sense -- there are skillsets that have synergies so that you can advance further in a basket of skills in a particular time than you would expect from considering them alone -- but the execution looks pretty bad, to me (although better than it was in Arms Companion, where they comprised perhaps the worst element of the worst RM Companion). They made sense in MERP as the equivalent of the RM adolescence packages, too, based on race/culture.
Maybe the problem is just that the RM skill-buy system (which I generally like, even the idea, in principle of the RMSS/FRP category/skill divide) wasn't designed with ease of conversion to a potentially large number of training packages across a wide number of professions. Also, people that see training packages as alternatives to professions (ie, classes), which are not about cost synergies but about personal qualities relating to what particular individuals can learn quickly versus what they can't, would be, in my opinion, misusing them.
Arguably, as well, some Training Package costs for vocational style TPs that are attained through an organisation, are lower than the sum of their parts in part because of an established teaching regime with teachers who are good at their job. That's a complicated road down which to go, however.