I’d be fine with tags in the digiKam DB including all of the above, but the values written to the images IPTC keywords field would have to be limited to the subkeys of what you listed above as type and content hierarchies. Here’s an example of one of my images: https://www.flickr.com/photos/tcgoetz/53265234720 you can see what I use for keywords in the info below the image.

My workflow also includes grouping similar images, using compare to grade them best to worst, marking the best with the pick flag, and rejecting the worst. I also group subimages of panoramics with the generated panoramic as the top of the stack and the pick image. didKam has a group feature but it seems different. Grouping in LR hides members of the group other than the pick image.