In the conlang I'm working on, there are a chunk of fairly new to me prosodic features that interact and that apparently are tied together a lot in natlang diachronic prosody analysis, but I can't really find any resource that seems to describe in any great detail how these things tie together.

1. qualitative stress (post-lexical pitch accent)
2. Heavy or overlong syllables are accented. I've been calling the high tone/glottalic/retracted ones stressed and the low tone/overlong/slightly murmured ones long.
3. "Stressed" syllables are grammatical. "Long" syllables are lexical. Long syllables win.

4. There's basically only one accented syllable per word (and an extra-long pause between words when accents are too close).
5. Stressed syllables can be realized as simply retracted and devoiced instead of fully glottalized (as ejective/implosive depending on voice), where q can be an allophone of ejective k or implosive g (which itself realizes as ejective devoiced in some conditions). The uvular g allophone is disappearing into q along with the implosive.

6. Stressed syllables are generally heavy, stealing the onset consonant of an adjacent syllable to give itself a coda. I used to not even bother marking syllables in citation forms because it depended entirely on stress where the syllable boundary occurred.
7. Long vowels are called long because there's only a two-way phonemic contrast (though diphthongs are pronounced in between), but they are pronounced at least twice as long as short vowels and sometimes a bit longer. They always have low tone. Short vowels or dipthongs never do except when receiving low tone spread from a long vowel.
8. Even not stressed, word-final consonants are generally partially or fully devoiced, often with no audible release and/or glottal stop reinforcement.
9. Stressed syllables seem to me to have started on word-final syllables and migrated to other positions when high/glottalized accent became grammatical.

So while it was great to narrow this down to pitch accent, in studying stod (wrong grapheme I know), pitch accent, moraic prosody on templatic morphology (both American polysynthetic types and Semitic), glottal reinforcement which often gives rise to simple register tone systems, and how all of this plays differently with short vs. long, back vs. front, and lowered vs. raised vowels, I keep trying to figure out which of these gave rise to which because it feels like I remove one of these interlocked features and the whole thing collapses.

Which isn't asking anyone to figure it out for me. But I'm gleaning bits and pieces of single natlang analyses of how one natlang's combined features might interlock and a lot of times, the description is very skimpy. Is there any good resource for understanding diachronic prosody with these kinds of systems in general? 

I feel like everything I'm studying says these things tend to reinforce each other, but there's little explanation of how or why or really how this glottalization complication fits into the register tone and moraic prosody. I find a good paper describing one phenomenon but only little bits on how it fits with one of the other two, but enough to know these three things do heavily interact in natlangs that have them.