I've been interested to notice that my interest in new developments in Obsidian has dropped right away.
Apart from the students & developers who appear to be the core of the user group, I ought to be a prime target, since I make a lot of notes, am an active writer, and most of what I write has been in .md or .txt for a long time. They see Obsidian as a PKM application, but much pre-existing K is stored in non markdown files. The freedom of feeling that I could forget about conversions and just work with data and files in original format has been liberating.
I'm a little surprised I didn't develop it before, as I'm sure many people have been doing it for years, but the concept of nested vaults has been key. (And, conceptually, the 'vault' is quite different in this context to 'folder'.) But it absolutely requires relevant files to be in the vault. It's data and file organisation by project rather than folder. It means a 'vault' can be copied and moved separately on to any device or shared with other people, with no impact on anything else. Links work, but then the file location has to be considered and managed separately. I don't mind having more than one copy of a file if necessary, but once a project is completed the vault is automatically subsumed into something larger and any issues with doubling can be dealt with then. I'm not keen on symlinks, but maybe that will change.
For tags, I'll use a mix of #tags in text files and windows meta fields, and the XY database approach. This does imply potential issues in some future decade but the theoretically more resilient approaches just come with too high a cost at this point.
And I'll save everything into the nested vaults. (I am aware that this may cause Obsidian itself to slow dramatically.) I'll use Obsidian where it works, and other programs and formats when they are better. I'm happy to type wiki-links and #tags into all text tiles, whatever their extension.
Search will give me #tags and [[backlinks]] and identify the existence or not of files for [[links]]. The greatest friction will arise from creating the linked files that do not already exxist. That's okay, I anticipate it taking time to establish the best search system and optimal way of creating those files. Obsidian has system for marking blocks and linking to headings; I've never use either and have always felt the block marker to be an ugly kludge. I doubt I will need a replacement, but think I will just use a timestamp if I do.
Tranclusions are another matter. They are, effectively, a property of Obsidin .md files. By continuing to use Obsidian, I won't lose anything I have but I don't have an alternative for other formats.