We spend about a third of our lives unconscious. A significant portion of that, we're having vivid, narratively structured experiences involving people we know, places that don't exist, and emotional arcs that feel completely real until the moment we wake up and they dissolve. The memory consolidation theory doesn't explain the narrative. The threat simulation theory doesn't explain why we dream about mundane things. The activation-synthesis model basically says it's noise, which feels like giving up. I'm not satisfied with any of these and I don't think the field is either.
That's consistent with some threat simulation and emotion regulation accounts, for what it's worth.
The memory consolidation evidence is actually stronger than the description suggests - there are fairly clean studies showing sleep disruption impairs consolidation in specific memory types. But you're right that it doesn't explain the narrative content. Memory consolidation would work fine without the subjective experience of a story.
That's the specific gap I find dissatisfying. Why does a memory consolidation process produce a first-person narrative with emotional stakes? The phenomenology seems like it should have an explanation but none of the functional theories touch it.
I had a recurring dream for 15 years that stopped the week I resolved something I hadn't consciously admitted was bothering me. I'm not a scientist but that seems like data.