"Centralization of mining due to slow propagation with bigger blocks" is mostly a strawman argument.
Even if the blocksize went up to 8MB with no increases in Internet speed,
you're talking about 8 seconds difference between an 8mbit connection and
a 16mbit connection. Compare to the 600 seconds required to solve a block
and you get 8/600 = .0133~. So that's a 1.3% advantage to the faster
miner. Quite dubious to say that would be a crushing competitive advantage
given that there are other factors involved in mining costs such as electricity,
gear, and operations.
Sry, I can't follow your numbers there with regards to block propagation. But mining centralization has already largely happened because of the economics of Bitcoin mining. If, however, someone wants to contribute with a basement space heater they can do so through the pool of their choosing with minimal network requirements to themselves.
Your last point is a good one. The chinese have cornered the market in large part due to artificially low electricity costs due to chinas centrally planned economy. If they lose out because they can't keep up with bandwidth and consistently gets stuck in the last 50th percentile when it comes to block propagation, then tough luck! If they can't compete then there is no reason for Bitcoin to keep them around. But there are ways around this problem, and the chinese miners know this, none of them are opposed to larger blocks.