I can't help but wonder if perhaps "dreaming" serves no function but is just a side-effect of our brain-cells shrinking (?)

There is some evidence that dreaming is a side effect of defragging the memories of the day into long-term storage format.

There are related learning benefits to sleeping, in particular for materials studied shortly before sleep.

http://healthysleep.med.harvard.edu/healthy/matters/benefits...

http://www.psywww.com/intropsych/ch06_memory/studying_before...

Or to 'entertain'/'occupy' the still operational brain-cells during the flushing period (like tv shows and movies during long flights or dialysis).

Brains are metabolically costly, if it were frivolous is probably wouldn't happen.

I'd wager it's more like keeping an engine idling because it's less expensive/dangerous than turning it all the way off and restarting it.

From what I've read, the brain partially 'compartmentalizes' it's communications at regular intervals, and dreaming is what happens when this process finishes: the different compartments start communicating, synthesizing multiple independent thought processes into a cohesive whole. (Which would explain why they can be seemingly random and also relevant of recent events.)