Cardano database artifacts on IPFS - MVP

## Why

The PoC validated that the aggregator can publish immutable file archives to IPFS and the client can retrieve and verify them. However, a single IPFS seed node operated by IOG still centralizes the hosting burden — the cost merely shifts from GCS egress to IPFS gateway bandwidth without reducing it.

The real cost reduction comes from **distributing the hosting across community volunteers**. Anyone running an IPFS node can pin Mithril immutable archives and serve them to other peers, effectively forming a community-powered CDN. Since Mithril certificates already guarantee the integrity of every immutable file archive through its signed Merkle root, volunteers don't need to be trusted — the cryptographic verification is built into the client regardless of where the data comes from.

The goal of this MVP is to deliver a production-ready IPFS distribution path where:
- The aggregator publishes immutable archives to both GCS (fallback) and IPFS (primary), with `ipfs://` URIs included in artifact metadata alongside existing `https://` URIs
- The client downloads from IPFS first (via HTTP gateway or local node) and falls back to GCS transparently
- A machine-readable pin list is published so community volunteers can subscribe and pin Mithril data on their own IPFS nodes
- Clients that download and verify archives can opt-in to re-seed them, creating a snowball effect where the swarm grows with adoption

This shifts the distribution model from centralized cloud hosting to a volunteer-driven peer-to-peer network, reducing IOG's infrastructure costs proportionally to community participation while improving availability and censorship resistance.