Good points, I was planning to slowly refactor and improve this block for a couple of reasons:

• in the last release I introduced a regression because one collection was a set (i.e lastEnvironments) while another one was a list (i.e resolvedEnv...) in the following comparison: if (lastEnvironments.size != resolvedEnvironments.size || lastEnvironments != resolvedEnvironments). It was my fault for not paying enough attention but I think the code is brittle, I had multiple issues in the pas around this part.
• toSet() even though it resulted in unique pairs of workspace&agent, hided the fact there are workspaces with two resource and same agent name. The initial distinctBy was applied only between the agents of the same resources. Somehow I ended up with this borked workspace and I never knew it was possible. In fact Blink suggests that neither issues should be possible (two agents with same name under the same resource or two agents with same name in two different resources)
• ahh you are right I forgot to add that comment back.
• in the end the pairs need to end-up as a list in toolbox so I thought working only with lists makes the code more reasadble.

TLDR; this is an attempt to fix a regression while at the same time to improve the readability and make some things obvious. I can just convert everything to sets and the fix would be much clear.

Ahhhh I see the distinct by is applied to the merge of all the workspace's resources now, rather than applied to each individual set of resources. Makes sense!