SHA is already quantum resistant. The problem is with public key encryption, i.e. bitcoin addresses.

The problem that Bitcoin relies on being hard (with regards to the proof of work, not addresses) is different from the one e.g. the standard PKI relies on being hard. It's not very likely the hashes would be broken by an advance in quantum computing; there are some sub-exponential speed ups that have been explored but nothing that would break the blockchain. It would be more akin to the movement from GPUs to ASICs.

why not implement post-quantum crypto now if there’s already a plan in place? this seems like a situation where “better late than never” isn’t true

The post-quantum algorithms haven't been around very long and as a result don't have the same level of confidence. Most post-quantum implementations that are in or nearing production use both existing methods (e.g. DH) and a post-quantum counterpart (e.g. New Hope) and combine them in a way that requires both to fail for there to be an issue (e.g. XORing the produced secrets together).

There is one launching check out theqrl.org