If you can afford it ($500+), I’d strongly recommend getting NVivo, MaxQDA, AtlasTi, or other QDA software that is designed for exactly this kind of task. You will want paragraph-level coding, reports by code, etc. etc. and all of the crazy hoop jumping that’s required to do a fraction of the tasks you describe in Scrivener will be trivial stuff that’s covered in the first chapter of the manual. Scrivener is a writing program not a qualitative data analysis program, and (having tried a similar approach briefly going to MaxQDA) it’s really not the ideal tool for the task. IMHO (as a professional anthropologist and teacher of qualitative research methods).
For example, rather than having your coding as colored styles with no way to assign multiple codes to the same passage, you’d be able to assign multiple codes and have them all displayed in the margin, then filter the document to display only passage with a given code or codes, or combinations (e.g. passages where people discuss money but not credit etc,). See e.g. maxqda.com/help-max12/how-t … nt-browser
If this sounds heavy-handed, it’s because for me, tasks that literally took days with a combination of pen and paper notebooks, word processor files , and a FileMaker database that indexed everything, I can now accomplish in seconds, No more cutting and pasting to gather all the notes on a given topic together - once the documents are coded, all the passages with a given code are instantly available.
If you’re attached to a university you might check if there is an institutional site license for any of the above.