[P1.5-S6-ARI-40] Refactor `GCPLogConnector` — configurable `resource_type_templates`

[bjridicodes]

Ticket: P1.5-S6-02

Type: Feature | Est: 1.5d

Goal: GCPLogConnector accepts a config dict of resource type → filter template list so new GCP service types can be added via conf.yaml without any Python code changes.

Scope:

• implementations/clusters/cloud/gcp/log_connector.py:
  • Add resource_type_templates: dict[str, list[str]] | None = None to constructor
  • Refactor _build_filter(): when resource_type_templates is set and the target resource type is found, use the template's filter clauses; otherwise fall back to the existing generic filter (backward compatible)
  • Template format:

    {
      "cloud_dataproc_cluster": ['resource.type="cloud_dataproc_cluster"'],
      "bigquery_dataset": ['resource.type="bigquery_dataset"', 'protoPayload.serviceName="bigquery.googleapis.com"'],
      "cloud_function": ['resource.type="cloud_function"'],
      "pubsub_subscription": ['resource.type="pubsub_subscription"'],
      "gcs_bucket": ['resource.type="gcs_bucket"'],
    }

• api/dependencies.py — get_agent2(): load resource_type_templates from gcp_resource_type_templates() config helper when constructing GCPLogConnector

Acceptance criteria:

• Existing Dataproc queries still work without resource_type_templates set (backward compatible)
• resource_type_templates={"bigquery_dataset": [...]} returns BigQuery logs from Cloud Logging
• Unit test: connector built with templates; assert _build_filter() output contains correct resource.type clause for a configured type
• Unit test: connector built without templates; assert existing generic filter behaviour unchanged