borg2: chunks index storage

[ThomasWaldmann]

Currently, it:

• involves CHUNKINDEX_HASH_SEED into the sha256 hash computation. this could be used to invalidate old indexes if we change stuff in borg.
• uses cache/chunks.{hash} as storage name.

It should:

• do a pure sha256 hash of the content. borgstore supports verifying storage files by sha256, but does not support a seed. maybe we can "kill" a now-invalid index of a previous borg version by some other means.
• use index/{hash} as storage name. the index is not a cache. while it can be rebuilt in case of emergency, that is slow and expensive, so index should be rather seen as an important, permanent part of a repository. also, just use the hash as the name, so borgstore's hash verification can be used in the same way as for other objects that have name = sha256(content).
• add the index/ namespace with correct nesting and permissions configuration (same as cache/?).