I was thinking “multihash” rather means the multiple hashing codes generated by different choices of hashing/pre-hashing algorithms, in order to make the content more distinguishable from different perspectives.
The purpose of the project is to handle one hash as a “string” that might be generated from multiple algorithm.
It is not that 1 hash-string has multiple hashes inside, but that it is self-describing what kind of algorithm has been used, and so the software can compare it easily.
Checkout http://multiformats.io