Students being graded by outside bodies would change the student-professor relationship radically (which is a further argument for such a reform, in my view). Today professors are both supposed to help students to learn and to judge their knowledge and their abilities. Under a system like the one I'm sketching, the professors would only have the former role.
Compare with driving school. In driving school me and my instructor worked together to maximize my chances of passing the test. The instructor didn't need to "grade" me (whatever that would mean in that context) to get me to show up on time, to do my homework, to do stuff I found boring, etc, because I knew it was in my best interest to do all these things if I wanted to pass the test. We co-operated to reach our joint goal. This contrasts starkly with the university educations, where professors and students have conflicting interests (nicely illustrated by some of your points).
Under this system, professors would essentially serve students to reach their goals. Students would presumably be much less subservient if the power of grading was removed from the professors. All this is to be welcomed, in my view.
So you are arguing that professors shouldn't test their students (the only testing that should exist is external). I vehemently disagree. You've assumed that testing is solely about evaluation - separate from learning - when in fact it is about both. Students learn better when tested, and teachers teach better when they understand how students are thinking.
In your driving school example - testing and learning are not separate. The instructor is in the car with you - he is constantly testing you and giving you advice on what to improve. Your system would ta...
One of many problems with the contemporary university system is that the same institutions that educate students also give them their degrees and grades. This obviously creates massive incentives for grade inflation and lowering of standards. Giving a thorough education requires hard work not only from students but also from the professors. In the absence of an independent body that tests that the students actually have learnt what they are supposed to have learnt, many professors spend as little time as possible at teaching, giving the students light workloads (something most of them of course happily accept). The faculty/student non-aggression pact is an apt term for this.
To see how absurd this system is, imagine that we would have the same system for drivers' licenses: that the driving schools that train prospective drivers also tested them and issued their drivers' licenses. In such a system, people would most probably chose the most lenient schools, leading to a lowering of standards. For fear of such a lowering of standards, prospective drivers are in many countries (I would guess universally but do not know that for sure) tested by government bodies.
Presumably, the main reason for this is that governments really care about the lowering of drivers' standards. Ensuring that all drivers are appropriately educated (i.e. is seen as very important. By contrast, the governments don't care that much about the lowering of academic standards. If they would, they would long ago have replaced a present grading/certification system with one where students are tested by independent bodies, rather than by the universities themselves.
This is all the more absurd given how much politicians in most countries talk about the importance of education. More often than not they talk about education, especially higher education, as a panacea to cure for all ills. However, if we look at the politicians' actions, rather than at their words, it doesn't seem like they actually do think it's quite as important as they say to ensure that the population is well-educated.
Changing the system for certifying students is important not the least in order to facilitate inventions in higher education. The present system discriminates in favour of traditional campus courses, which are both expensive and fail to teach the students as much as they should. I'm not saying that online courses, and other non-standard courses, are necessarily better or more cost-effective, but they should get the chance to prove that they are.
The system is of course hard to change, since there are lots of vested interests that don't want it to change. This is nicely illustrated by the reactions to a small baby-step towards the system that I'm envisioning that OECD is presently trying to take. Financial Times (which has a paywall, unfortunately) reports that OECD are attempting to introduce Pisa-style tests to compare students from higher education institutions around the world. Third year students would be tested on critical thinking, analytical reasoning, problem solving and written communcation. There would also be discipline-specific trials for economics and engineering.
These attempts have, however, not progressed because of resistance from some universities and member countries. OECD says that the resistance often comes from "the most prestigious institutions, because they have very little to win...and a lot to lose". In contrast, "the greatest supporters are the ones that add the greatest value...many of the second-tier institutes are actually a lot better and they're very keen to get on a level playing field."
I figure that if OECD get enough universities on board, they could start implementing the system without the obstructing top universities. They could also allow students from those universities to take the tests independently. If employers started taking these tests seriously, students would have every reason to take them even if their universities haven't joined. Slowly, these presumably more objective tests, or others like them, would become more important at the cost of the universities' inflated grades. People often try to change institutions or systems directly, but sometimes it is more efficient to build alternative systems, show that their useful to the relevant actors, and start out-competing the dominant system (as discussed in these comments).