For the input of the ripe-md160 function the publickey (compressed and uncompressed makes no difference) will get hashed with sha256.
So there can be collisions for one address (conclusion) in the range of 2^96, according to the birthday paradox.
Just to mention something significant: It's not the fact that the public key gets hashed with SHA256, which will give a 256-bit number as a result, the reason why there on average 2
160 different combinations. It's the range of public keys that is ALSO around 2
256. If the range of public keys was not this, but something like [1, 10], there could only be 10 different SHA256 hashes; not ~2
256.
2. How many private keys are in the range with the same Compressed address?
You mean how many private keys give the same compressed address? As said, approximately 2
256 / 2
160 = 2
96. That's true for uncompressed addresses too.