I've been thinking about plant intelligence for over a decade (and have a neuroscience-related Ph.D.), yet never took it down the path of computation efficiency/time/energy tradeoffs as you have done. Thank you for this new perspective, including its obvious conclusion:
> trees can afford to be very, very smart.
Hopefully I can use some of my perspective to address your follow-on question:
> What could a smarter tree do to outcompete rivals?
This may represent a limited understanding of Darwinian evolution.
Trees have an economy, sharing resources with each other and with other living things.
Tree also are (more or less) fixed in place, thus these economic games are repeated games with known participants.
They thus have a vested interest in maximizing the wealth of the community, rather than their individual wealth. They may (and studies suggest they do) have a preference for supporting their offspring over other trees, but not to the point of killing the community their offspring will also depend on.
Where I am going with this. Repeated economic games are not zero-sum; the pie can get bigger. "Competition" in this setting does not look like competition in a zero-sum "I have to beat the other guy" game.
What trees can do with their intelligence is learn how to build long-term stable and nuturing communities which generate substantial wealth for many members. And that this is a logical local (perhaps global) optimium.