animedex.cache
Local SQLite cache for animedex backend responses.
The cache is keyed by (backend, request_signature) and stores
arbitrary bytes per row, with a per-row TTL. Backend modules use it
to keep the upstream call rate well under the P1 caps from
plans/02 and to make --no-cache and --cache <ttl> flags
straightforward to honour.
The submodule split is deliberately thin:
animedex.cache.sqlite- the storage engine and the default TTL table.
The cache layer never imports any backend module so circular-import issues do not arise.