Emit typed bootstrap skip outcomes

[riyazsh]

Summary

• add typed SkipResource metadata and optional NDJSON details for actionable skip outcomes
• emit destination_metric_missing for metrics_metadata, metric_percentiles, and metric_tag_configurations with metric_name and operation details
• emit integration_pipeline_bootstrap_required for logs_pipelines bootstrap gaps with pipeline_name and ddsource details

Context

Paxos DDR needs sync-cli to distinguish terminal/no-op skips from skips that require managed-sync bootstrap. PR #669 covers metric tag configuration behavior, but metrics_metadata and related metric resources still serialized missing-destination metrics as failure_class=unknown, leaving managed-sync without a clean signal.

This keeps the existing HTTP failure taxonomy intact and only adds failure_class/details when a resource model deliberately raises a typed SkipResource.

Testing

• .tox/py312/bin/pytest -q tests/unit/test_failure_class.py tests/unit/test_sync_report.py tests/unit/test_metrics_metadata.py tests/unit/test_metric_percentiles.py tests/unit/test_metric_tag_configurations.py tests/unit/test_logs_pipelines.py
• .tox/py312/bin/black --line-length 120 --check datadog_sync/utils/resource_utils.py datadog_sync/utils/sync_report.py datadog_sync/utils/resources_handler.py datadog_sync/model/metrics_metadata.py datadog_sync/model/metric_percentiles.py datadog_sync/model/metric_tag_configurations.py datadog_sync/model/logs_pipelines.py tests/unit/test_failure_class.py tests/unit/test_metrics_metadata.py tests/unit/test_metric_percentiles.py tests/unit/test_metric_tag_configurations.py tests/unit/test_logs_pipelines.py
• git diff --check