chore(datadog): resolve compare/data-readiness errors from release dashboard batch

## Summary
Post-release follow-up for Datadog compare/data-readiness failures in the release dashboard batch.

## Evidence
- Datadog live ES|QL acceptance was clean: 111/111 queries ok.
- `obs-migrate compare` returned 257 STRUCTURAL and 37 ERROR rows.
- Top errors were missing target fields/data-readiness gaps, including examples around `activemq_queue_*`, `mysql_innodb_buffer_pool*`, `node_address`, `table`, `columnfamily`, and `jmx_gc_*`.
- Sample-data seed partially failed: 254,505 docs ingested, 11,913 ingest errors.

## Acceptance criteria
- Seed data contract covers the Datadog integration fields needed by this batch, or reports unsupported fields explicitly.
- Compare no longer emits ERROR for data-readiness cases that can be seeded.
- Remaining structural-only Datadog checks are clearly labeled as not numeric proof.

Start by tracing the sample-data seed and `obs-migrate compare` paths that produce the reported STRUCTURAL and ERROR rows. Use the listed missing fields and seed-error counts to identify data-readiness gaps, then verify that seedable cases no longer emit ERROR and structural-only checks are labeled as non-numeric proof.

Written by the indexing model from the issue text.