feat(subgraph): index Warm Storage datasets per payer

### Description

The Warm Storage service page needs, per dataset owned by the connected payer: dataset id, provider, size, last write (last piece added), and the rail(s) funding it. None of that is queryable from the payer side today. This gates the dataset table (#463), triage queue (#465), and inactivity email (#467).

Index FWSS dataset creation, piece added, and rail association events into a `DataSet` entity keyed by payer, with `lastWriteAt`, `totalSize`, `railId`, and `status`. Proving health stays out of the subgraph; the UI links to PDP Explorer per dataset.

### Additional context

Parent: #460. Event list and entity shape to be confirmed against the FWSS contract ABI. foc-observer already indexes these events and can be used to cross-check counts.

Merges alone: additive schema only, no existing entity or field changes, so the current explorer keeps working against the new deployment.

Start by confirming the FWSS contract ABI event list and the proposed DataSet entity shape, then compare expected events and counts with foc-observer. Define the additive schema and indexing for dataset creation, piece-added, and rail-association events, with payer ownership and the listed fields. Done means the new deployment exposes the required per-payer dataset data without changing existing entities or fields.

Written by the indexing model from the issue text.