Descriptive Problem Summary:
Variation in cpu usage (between high and low numbers) leads to client-side lag and stuttering, EVEN IF WE DO NOT ENTER OVERTIME (seemingly dropped visual information from byond? but it's hard to tell, it doesn't seem to impact glide animations I don't think).
I am pretty sure this occurs whether or not the client and server are on the same machine, I've seen it happen in testing remotely but due to hosting issues I can't really check right now. It replicates even if dreamseeker and dreamdeamon are pinned to different cpu cores

The video below demonstrates the issue.
https://file.house/vpG7WJ2cpcNm7SXxMW7Qgg==.mp4

Numbered Steps to Reproduce Problem:
1. Setup code in world/Tick() that runs a while loop until X amount of cpu is consumed.
2. Block it behind a prob() check.
3. Set the cpu consumption amount to like 90, and the probability to 50.
4. Move around.

Code Snippet (if applicable) to Reproduce Problem:
https://github.com/LemonInTheDark/basic-test-project/tree/ cpu_bending
I've included my cpu deaveraging code in order to provide more precise info about current/min/max cpu (this version has some floating point percision errors I think but it's not a huge issue in this timescale).

Expected Results:
No visible lag as we have not entered into overtime

Actual Results:
Visible stuttering, on /tg/ this looks like delays between finishing a move to a turf and starting a move to the next turf.

Does the problem occur:
Every time? Or how often? It's seemingly random, but it's consistent given time.
In other games? Yes
In other user accounts? Yes, I've tested with a partner on a remote machine and we both saw the same behavior
On other computers? ''

When does the problem NOT occur?
When cpu usage is stable

Did the problem NOT occur in any earlier versions? If so, what was the last version that worked? (Visit http://www.byond.com/download/build to download old versions for testing.)

Workarounds:
IDK man manually use all available cpu (minus maptick?) to avoid large variations??