There is absolutly nothing special about this case, or a collision or what soever.
It is simply a different way of generating a bitcoin address. Say you create a "type 3" multisig address. One would create a "redeemscript", which is for a 2-of-3 example multisig this:
This redeemscript is then converted into a bitcoin address:
1.) sha256 of the redeemscript: 8342fc1facef672c9b6c96b8dc79ac7ff2bfbb6babe444d6abf40597f19e12c5
2.) RipeMD160 of the redeemscript: bc01acb41044f23a1eccaeb112e3a6021c3ab844
3.) base58 encoded -> P2SH address: 3Jq6xxnjXbB5BxFLfLidWmaunKBBSstJYn
This is a "type 3" address based on a redeem script.
When now using the RipeMD160 hash, and putting in the step 4 of the normal address generation, you receive a normal bitcoin address (as I don't have the priv key or the public key, neither the SHA256, I skip the first three steps of a bitcoin adress generation task):
0 - Private ECDSA Key
1 - Public ECDSA Key
2 - SHA-256 hash of 1
and then follow these steps to create the final bitcoin address:
3 - RIPEMD-160 Hash of 2
4 - Adding network bytes to 3
5 - SHA-256 hash of 4
6 - SHA-256 hash of 5
7 - First four bytes of 6
8 - Adding 7 at the end of 4
9 - Base58 encoding of 8
This creates me the bitcoin address 1J963RJHygrh6nYuYF4369DydntTtG3YBy, and has the same RipeMD160 hash.
So you can repeat this step with any "type 3" multisig address or its RipeMD160 hash. Nothing special ...