refactor: Remove the availability oracle

[LHerskind]

With changes planned for progressing to using blobs, we do not really need the availability oracle as it is.

I suggest that we get rid ahead of the chance since that would also aling the code path towards something that will look more like the blobs.

Namely, currently it is possible to publish the data in one call, and then a separate propose a block. Related to #8452 it would make sense to entirely remove that separate flow as it is not intended to be used with blobs anyway.

Our intended use is that the propose will also publish the data, so moving to that now should make the transition simpler.

At the same time, it should provide us with a simpler archiver that skips a lot of the "double" work it is currently doing, because it is fetching metadata and block content in separate fetchers, but they would always be provided at the same time.