Tethered power + laser cutting would solve the weight issue and the don't-drop-a-saw issue. Not sure how thick the branches are that need to be severed, though.
But drone solutions seem much more efficient, especially given how long it takes to set up the robot featured here (15 mins per tree). This time would undoubtedly be reduced in production units, but variability in trunks will always complicate things. With a drone, it doesn't matter.
A drone-carried laser powerful enough to cut branches that thick seems like a pretty big ask, not to mention the accidents when the laser shoots past the branch! Dropping a drone that heavy would be dangerous whether or not it had a saw blade.
I think you could get the setup down to 30 seconds with parts that clip tightly together, e.g. how folding bikes work. The controls seem like a bigger problem, as automating that for the wide variety of natural trees, conditions, and lighting seems challenging. Having it crawl to the next tree and wrap itself around it to climb would be pretty neat too.
This is why you'd do a drone. It's a good learning experience directly to industry.
A tree climbing robot is way more abstract. Which might be good depending on the degree. But you'll never work on tree climbing robots in real life.
Neither will ever be used by farmers to pick coconuts, so it depends on what you want.