I don't know why RIPEMD160 or HASH160 even chosen in first place.
Probably they just picked something, which was already implemented. However, I guess some people simply don't know about collisions. Existing double SHA-256 block headers have around 80 leading zero bits, for example: 00000000000000000000e081e5715f5301b9fdcc917a1a6accfd011e79cee225
Maybe for RIPEMD-160, it will take some time to get there, to build proper ASICs, and to reach that kind of hashes, but still: if you have 160-bit hashes, and half of that hash will be set to zero, then miners will have a choice: mine the next block, or try to find a collision. And then, by finding collisions, it is possible to disrupt the network, by serving one value to one group of nodes, and another value to another group of nodes. Good luck doing Initial Blockchain Download later, when nodes won't know, which value is the original one, if both will give you the same hash.
If they actually do it and manage to split the chain, i will prepare my popcorn and wait for someone to create TX that include controversial/questionable arbitrary data just to piss them off.
When many people will suddenly receive the same amount of BIP-110 coins, then the first thing they could try, is selling these coins. But if that would be unsuccessful, because of lack of support from exchanges, then of course, many users may decide to burn everything into fees, just to upload some data, and show BIP-110 supporters, that their fix won't stop the spam.