Refactor storage cache

[germag]

This is the first part of a larger refactor. We have some issues that need to be fixed:

• Early on I made the mistake to store the disk images in the root of the cache directory, this will be fixed in the future.
• The cache/storage logic is all over the place, this makes it difficult to access with the right locks. I can make this works without locks, just using atomic syscalls (i.e, rename, *xttr), but those are guaranteed to be atomic only on posix systems. AFAIK Windows doesn't provide atomic Fs operations (the use of TxF is discouraged).
• We store all the metadata in a single json object, this also makes difficult to access it for different commands the just require partial info.
• Modeling the Vm as an object and binding all the operation to that object (i.e, Run(), IsRunning(), Delete(), etc) cornered us making very difficult to add new features, like the BIB support, rebuilding the image if the disk size changes, etc.

I'm marking this PR as draft because depends on PR #56

[germag]

@sourcery-ai review