Add an integrity check to blob list

[rikettsie]

Context

rdrop blob list shows all imported blobs and their transfer tickets, but currently does not verify that the underlying files are still present and readable on disk. This implies a blob entry becomes silently stale when the original file is deleted or moved after import, producing a ticket that will fail at transfer time (!).

Goal

Show broken blob entries at blob list time so the user knows to remove and re-import them. Expose the check as a reusable function so the daemon can apply it before serving a transfer and anywhere else it's useful before manipulating a blob.

Acceptance criteria

• A is_blob_readable(hash) -> bool function is exposed on Node to check whether a blob's backing file is present and readable;
• This check is "best-effort": any I/O failure accessing the file is valid to mark the blob as "broken";
• For directory blobs, the check verifies that every file in the collection is present and readable (a single missing or unreadable file marks the whole blob as broken);
• rdrop blob list calls this function for each entry and marks broken blobs in the output with [broken];
• A hint is printed when at least one broken blob is found, suggesting rdrop blob remove <hash> to clean up the registry;
• The daemon rejects any transfer requests for a broken blob with a clear error to the requesting peer, rather than failing mid-transfer;
• Healthy blobs are unaffected in output and behaviour.

Out of scope

• Automatic removal of broken entries;
• Deep content verification (re-hashing the file against the stored BLAKE3 hash);
• Watching for file changes at runtime.