These kinds of hash functions are actually /designed/ to be slow (so you can't use brute force)
Not so much. I think they are designed to have good mixing, be impractcally hard to reverse, and have other properties. This just makes them slower than hash functions generally used in hash tables. My understanding is that both MD5 and SHA were designed to be efficient under these constraints. bcrypt is designed to be slow, and can beat other cryptographic hashes by ridiculous margins. (Parameterized, actually)
Not so much. I think they are designed to have good mixing, be impractcally hard to reverse, and have other properties. This just makes them slower than hash functions generally used in hash tables. My understanding is that both MD5 and SHA were designed to be efficient under these constraints. bcrypt is designed to be slow, and can beat other cryptographic hashes by ridiculous margins. (Parameterized, actually)