…elds (#23543)

* Fix snmp generate-profile-from-mibs to use modern metric_tags symbol field

The ddev meta snmp generate-profile-from-mibs command was emitting
table metric tags using the deprecated `column` field, which the SNMP
Profile Manager rejects on upload. Switch to the modern `symbol` field
per the MetricTagConfig schema in datadog-agent's profiledefinition.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Rename changelog fragment to match PR number

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Accept `symbol` in TableColumnHasTagValidator

The schema in datadog-agent's profiledefinition deprecated `column` in
favour of `symbol` for table metric tags. Extend the local
validate-profile validator so profiles using the modern `symbol` field
are not flagged as missing tags.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Pin GitHub Actions to full vX.Y.Z tags in Renovate

* Limit Renovate to action depType only

Renovate's github-actions manager also extracts runner versions (runs-on), action input versions (uses-with, e.g. node-version), and container/service images. Scope the action-only rule to matchDepTypes:[action] and explicitly disable every other depType so runner/input/container bumps stay manual.

* Allow Renovate to bump uses-with inputs in same group

Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>

* Update .yaml workflows in update-python-config

The CI file glob only matched .yml, missing workflows like
claim-pypi-name.yaml and dependency-wheel-promotion.yaml, which is
why some workflows still pinned older Python versions while the rest
of the repo had moved on.

* Add changelog entry

* Fix on_error type to OrchestratorHookError and remove None return types from tests

* Add changelog

* Make on_error and _apply_error_policy generic over wrapped error type

Codex flagged that _apply_error_policy is also called from _task_wrapper
with MessageProcessingError and ProcessorHookError, so narrowing it to
OrchestratorHookError broke mypy. Make it generic on E: Exception and
turn ErrorHandler into a generic alias so each scope keeps the
relationship between the wrapped-error type and its handler. Tighten
BaseProcessor.on_error to MessageProcessingError | ProcessorHookError to
match its actual contract.

Keep the legacy schema detail connection open for the duration of each database schema collection instead of reconnecting once per table.