feat: allocating inline IPLD multiformats

[mikeal]

We've talked about this informally for a while, I'd like to start experimenting with it but I want to make sure we get low numbers allocated to basic inline IPLD types.

[mikeal]

We may also want to allocate uint8, uint16, and uint32 multiformats for when we need to do fixed length number encodings.

[rvagg]

bit of a can of worms here if we start adding all the possible types in the single byte range - various string encodings, float encodings, all the ints and their endianness forms; so maybe this should be done in the two-byte range so we can add whatever people want?

[vmx]

I don't know what inline IPLD is used for. Is the idea to use it inside CIDs?

I'm with @rvagg here. Those experimental things belong in the >=2 bytes range.