Cleanup instances of manual resolver construction

[magik6k]

Part of #5491

Given how messy and all-over-the-place this PR is, I'd rather keep it small. This only cleans up places which constructed their own resolver. I'll take care of the other parts, which use node.Resolver in next PR

[schomatis]

Could you provide me some pointers about this PR that would help me review it?

[magik6k]

The idea (mentioned in #5491) is that we want to start using the coreapi resolver basically everywhere.

The resolver was discussed in #4666 / #4672, the TL;DR is that it decides which resolver to use based on the prefix, so that /ipfs/ paths will always resolve with the unixfs resolver (handling hamt) and /ipld/ with the simple resolver (which basically uses IPLD paths). It also handles /ipns/

[schomatis]

This code seems to be repeating the interfaces already defined in go-unixfs:

https://github.com/ipfs/go-unixfs/blob/master/io/dagreader.go#L22-L37

(Or maybe it's the other way around and UnixFS is duplicating these ones, either way, is it possible for one to reference the other?)

[magik6k]

That will be doable once the coreapi interface is extracted, which I hope will happen fairly soon - for progress see #4498 (I want to have pubsub/swarm/unixfs done before extracting the interface)