That link is very out of date. The most recent fork I am aware of is at block 772,981, which was around 2 weeks ago. There was a competing block with the hash 0000000000000000000682990a0dae862b48e0451d619938215dd47ed9560200 mined by Foundry. Usually we see on average around one such event a month.

I completely agree, with bitcoin as it is just now. But if you are talking about inflating the block size to several gigabytes or more, then such stale blocks become significant more frequent, and the chains of stale blocks also become longer.

It's actually 4 GB.

Because it takes longer to download, verify, and then transmit that block to other nodes. It can take tens of minutes for such a block to make its way across the entire network, as opposed to the few seconds as happens in bitcoin at present. During those minutes, all other miners will continue to work on top of the previous block, and therefore have extra time to find a solution. If they find a solution by the time the other block finally arrives at them, they'll simply ignore it and keep trying to build on their own block, causing a growing re-org.

Let's call you miner A, and call me miner B. We are both mining on top of the same block. You find a block, which we will call A1, which you transmit. It take minutes to reach me, and in that time, I find block B1, at the same height as A1. You are now mining on top of A1, and I'm mining on top of B1. I find B2, and transmit it. By the time it reaches you, you have found A2, and some other miner has also found C2. Each of us keeps mining on top of our own blocks, because the time delay in transmitting such oversize and bloated blocks means it takes a long period of time until someone finally gets ahead enough to win the race, resulting in a re-org many blocks deep.

Here's another link explaining this: https://bitcoin.stackexchange.com/questions/86169/why-do-large-blocks-increase-the-probability-of-chain-reorgs