[BUG] High CPU usage during thinking generation

[allo-]

Describe the bug
Zoo Code maxes out one CPU core while streaming text (like LLM thinking). I think it may use inefficient updates to the text in the sidebar causing costly rerendering of the full section or something similar and it would be worth looking in which rendering operations are the most efficient

To Reproduce
Steps to reproduce the behavior:

1. Let Zoo work on any task using a thinking model (e.g. heavy thinkers like Qwen3.8)
2. Observe 100-150% CPU usage from the code/codium processes

Expected behavior
Negligible CPU usage while streaming LLM output

What version of zoo are you running
3.60.0 (a62a652) (codium marketplace version)

[justcallmedots]

I can confirm this does happen a lot and i hope they fix the issue. I have a ryzen 7 and 3070 TI and 16 gb of ram but it uses too much CPU.
I hope they can fix this and fast.