FWIW since my last post this is what I've done:
1. Setup my pdf library in Zotero. However, I don't store the pdf in the Zotero database, My pdf's are stored in a cloud folder and I link to that folder in Zotero.
2. The pdf's are in one folder (no subfolders). In Zotero I ended up using a fairly 'flat' folder/group hierarchy with tags. I had not thought through using tags, and ended up with chaos; too many tags to be useful. Rather than redo the tags (which I was going to do) and having learned that Zotero tags are not actually 'open-meta' tags that the file system uses, I decided to forget the 'redo'.
3. I also installed DevonThink. Rather than import my pdf's into DT, I simply index the pdf's in the cloud folder. In this way I get the search and relational power of DT w/o losing the ability for other programs to also use the pdf's.
As I was thinking about tagging the pdf's in DT, I also learned that DT tags are not 'open-meta'. DT can use 'open-meta' but when you tag in DT they are only for DT. Regardless, DT's real benefit is its indexing and searching features. The tags probably aren't that useful in DT anyway.
It is my understanding that OSX Mavericks will fully implement 'open meta'. And if that turns out to be true, I would definitely want to use 'open-meta' tags and not some programs proprietary tagging system.
As for my RSS feeds, I experimented running them in DT, but it seemed clumsy in comparison to Reeder or Feedly. In the end I still use Reeder; and setup Feedly as a backup to the Google Reader aggregator. It looks like Reeder will hook to Feedly so I'm good either way. The entries I would normally 'start' I just clip to DT.
** If any of this is in error, hopefully someone will step in to correct it.
Lastly, I'm using nvALT for notes. It fully supports Markdown and has a fantastic indexing system of its own. All my notes go into a cloud folder.