Grouping resource collection to save CPU

In the original game, when a lot of workers all work at the same resource tile they make frequent collection calls (C)
time-------->
x---C1--C2---x---C1--C2---x---C1--C2---x

C1 stands for "call of worker 1", likewise C2, traits are frames, x is the nth frame

When the resource collection is off-screen (i.e. the player doesn't watch), we could "group" resource collection calls from miners, lumberjacks and others in a single one, to save CPU, so that, say every nth frame, the resource tile value is resource -= (number of workers) * (resource taken by each worker). As a worker is full, the resource (and the rate) diminishes by a worker until he comes back to work.

Something like:
time-------->
B{1, 2}----------B{1, 2}----------B{1, 2}---G1------B{2}---R1------B{1, 2}

• B{1, 2} is "big call {worker 1, worker 2}" (B is at frame nth),
• G1 is when "worker 1 goes",
• B{2} is big call with worker 2 only,
• R1 is when "returns 1"

The resources taken by each worker would be updated syncronously in the same nth frame, with a delay to take into account the time workers actually began working at the tile (so they're not full all at the same time - wouldn't be faithful of the original game)

The issue does not name any files, tests, or entry points. Start by locating the worker and resource-collection systems, then measure off-screen collection CPU cost before deciding whether grouped calls preserve worker timing and resource depletion. Done should include an implemented approach with evidence that behavior remains faithful and CPU usage improves.