fix(celerdata): collect all per-db starrocks_fe_table_num series

[jaogoy]

Why I'm doing this

StarRocks FE exposes starrocks_fe_table_num interleaved with starrocks_fe_db_size_bytes (one pair per database) under a single # TYPE ... gauge line, which violates the Prometheus text exposition rule that all samples of a metric family must be contiguous.

The prometheus_client streaming parser used by OpenMetrics V2 therefore splits the metric into separate families: only the first carries the gauge type while the rest are typed unknown. OpenMetricsBaseCheckV2 drops the unknown families via skip_native_metric, so only the first per-db series (db_name=information_schema) is collected and every other database silently disappears.

What I'm doing

Pin starrocks_fe_table_num to the gauge type in METRIC_MAP, using the same dict form already applied to tablet_num / thread_pool. An explicitly typed metric is handled by the gauge transformer instead of the default native path, which bypasses the unknown-type skip and lets every per-db series through. The interleaving itself is a StarRocks-side bug; pinning the type fixes it on the integration side without waiting for a StarRocks release.

Verified on a live Agent: celerdata.fe.table_num now reports all per-db series (information_schema, _statistics_, sys, plus user databases) instead of only information_schema.

Fixes #2854.

---

Notes for reviewers

• Depends on [CelerData] Add slow_lock_held_time_ms and and slow_lock_wait_time_ms metrics metrics #3021 (slow_lock metrics, bumps to 1.3.0). Please merge that first; I will rebase this onto master afterward, at which point it becomes 1.3.1. It is versioned 1.2.2 here because it currently branches off master (1.2.1).
• conf.yaml.example and config_models/ are regenerated via ddev validate ... --sync to clear pre-existing drift from the shared OpenMetrics config template (not specific to this change).

[datadog-prod-us1-5]

✨ Fix all issues with BitsAI

⚠️ Warnings

🚦 2 Pipeline jobs failed

PR | test / check     

PR | test / test-minimum-base-package (linux, ubuntu-22.04, celerdata, celerdata (py3.13), py3.13) / minimum-base-package-celerdata (py3.13)-py3.13     

See error

Failed to build `ddtrace==2.10.6`. Error running `setuptools.build_meta.build_wheel`: No module named 'pkg_resources'

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 1f57452 | Docs | Datadog PR Page | Give us feedback!}