The answer to any question about "can learn" has to reduce to the conclusion that anybody can learn any language (as a child), and there really isn't a solid basis for theorizing about how adult language learning of one language might be facilitated by previously knowing another, apart from the obvious fact that Spanish speakers have an easier time learning Portuguese than English speakers do. Likewise, any language of a given "type" can eventually evolve into a language of another "type".
The question could be framed as synchronic entailments, for example, you might posit that any language which allows syllable-final obstruents will also allow syllable-final nasals. From that perspective, any such hierarhy can't be strictly ordered (or, the meaning needs to be clarified). The basic division is into properties of the onset, and properties of the coda. Some languages allow onsetless syllables and others don't; some languages allow complex onsets, and others don't. There is also a hierarchy of possible onset clusters, related to the notion of sonority, where CG is the most ubiquitous cluster type and stop clusters are the least common. Conditions on the onset are generally independent of conditions on codas (e,g, proto-Slavic is assumed to have substantial onset clustering and no closed syllables); however, from a statistical POV, a language with rich onset clustering usually allows coda consonants, and a language with level-sonority onset clusters usually allows codas.