node: warn, and name both node ids, when no block ingestion runs

[cargopete]

When [chains] ingestor names a node that is not running, nothing ingests blocks. The chain head stops advancing and every deployment sits at its current block reporting healthy, so nothing reports a problem anywhere. graph-node already logs this at INFO, but the line names only the configured ingestor and not the node's own id, so it reads as a correct statement of intent rather than a mismatch.

This caught an indexer upgrading 0.41.1 to 0.45.0, crossing #6447 where the field became load-bearing. Their config said index-node-0; the node ran as index_node_0. Both are valid node ids so validation passes, and the log line said "Not running block ingestion, ingestor is index-node-0", which reads as correct. It took two and three quarter hours to find.

This raises the line to WARN and names both ids, so the mismatch is visible in the line that already exists. Behaviour is unchanged, and config::tests::is_block_ingestor already covers the logic.

A node deliberately not the ingestor will now warn once at startup. If you would rather that stayed INFO, I will drop the level change and keep the message.