Review the config API

[igaw]

    > Ideally we would validate against the schema.

After some discussion during ALPSS we should aim to move all the config reading parts to the fronted. The necessary data structs available to move the config data to the library.

Originally posted by @igaw in #491 (comment)

[igaw]

I've looked into it a bit more into it and figured out that it's probably not worth to do this. The tests cases are using the nvme_dump_tree interface.

Though one thing we should do is check if the nvme_config* API could be improved, e.g. it looks like we could drop nvme_scan.