Yes, been there since a long time, but it is becoming more pressing.
Discussion
I feel a little like a positive goal of "move to having thread X own the block tip info and mempool and communicate that to other threads via a message queue/atomics" or similar might be better than a negative goal of "use less cs_main"...