I don't know any annotated die photos, or any other analysis of this chip, but there are some (fairly low resolution) photos out there, so in the spirit of opening a discussion, I thought I'd post this. Don't take this as gospel truth: I'm not an expert at this. I've spent a while reading about chip design, but have no practical experience, and this is a long way out of my comfort zone, but... the best die photo of Agnus I can find is a long way from being good quality, but here's my best guess at the structures I can identify there:
What I'm reasonably confident about: the section highlighted in red. There are either 12 nearly-identical or 24 mirrored units in this array. Given the unusual number, and that Agnus has 24 DMA controllers, this is almost certainly what these are.
I'm not so confident about the section highlighted blue, but this is my best guess for where Copper is. It's certainly a complex section of the chip with a high degree of feedback, and looks similar to other processor control circuits I've looked at.
I'm even less confident about the green section, but that's my best guess at the location of the blitter. It's about the size I'd expect the blitter to be, and has a degree of regularity that you'd expect from a circuit of that nature (a lot of similar functions handled with different types of shift and mask patterns, and so on). I'd be more confident, however, if it wasn't so far from the DMA channels that I'd expect it to be interacting with quite heavily, so I'd definitely bow to anybody else's reasoned opinions on where else the blitter might be.