Your question seems to contain a number of false presuppositions which could be repaired by rewording. One is the distinction between absolute tendencies and implicational universals – the problem is the concept of an absolute tendency ("tendency" means that both outcomes happen, and one is more frequent, which contradicts the assumption of being absolute). The second is that properties can be manifested in a particular arrangement of vowel letters. Properties can be manifested in a language, and a language can have a particular collection of vowels, but tendencies are only latent in abstract vowel systems, so not all languages with the vowel system /a e o i u/ manifest the same properties.
One thing that you can do with the vowel system of a language that makes sense is ask whether there is any reason for that vowel system as opposed to come other vowel system. For example, if you have /i u a/, you could ask why the language doesn't instead have /i u a ə/. The typical answer is that the vowels [a] and [ə] are acoustically too similar, and it is hard to tell them apart. That in a nutshell is the prevailing explanatory force behind most questions about vowel systems.
Some of these supposed tendencies turn out to be tendencies about linguists, not languages. The real reason why there are no languages where all vowels are long is that length is not an absolute property, so however long a vowel is, a linguist will always call it a short vowel unless there is a contrasting vowel type that can be called long.
There are two main approaches to handling phonetic tendencies in language. One is the approach of Optimality Theory, which says that people are genetically encoded with specific prohibitions, for example there is a tendency that forbids nasal vowels, symbolized as *Ṽ, and language with no nasal vowels obey that constraint. Languages like French which have nasal vowels ignore the constraint. The set of constraints is universally predetermined, and by stipulating that there is no constraint forbidding oral vowels (in all contexts -- there is one forbidding oral vowels next to nasal consonants), OT can express the generalization that no language has only nasal vowels.
The alternative approach is the historical-evolutionary approach, where attention is paid to the phonetic structure of languages and inventory tendencies are attributed to different probabilities of people learning a given distinction, in the usual sub-optimal context where language is actually learned. For instance, you don't usually find vowel systems where all the vowels are significantly clustered around the center of the vowel space, instead they tend to maximize perceptual distinctness (Liljencrantz & Lindblom 1972 and much literature since). Nasalization in vowels does that sub-optimal thing -- it makes vowels acoustically closer to each other. Languages can tolerate vowel systems like /i e u o a ĩ ẽ ũ õ ã/ since ultimately people do have a remarkable ability to perceive subtle distinctions. But people also give in to perceptual expediency, and you often find that the number of nasal vowels is reduced relative to the oral set, because nasal vowels are the ones at a perceptual disadvantage. There would never be an advantage to having a set of vowels that are all nasal, hence ordinary linguistic evolutionary tendencies converge on the result that vowel systems are not all-nasal.
