ref(preprod): Fix build e2e time for reruns (#112527)

Pass a `triggered_at` timestamp when dispatching the size analysis
comparison task, so the `preprod.size_analysis.compare.results_e2e`
metric measures from dispatch time instead of `artifact.date_added`.
This fixes inflated e2e durations on re-runs where the artifact was
created much earlier.

Author: NicoHinderling

src/sentry/preprod/size_analysis/tasks.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import logging
+from datetime import datetime
 from io import BytesIO
 from typing import Any
 
@@ -53,6 +54,7 @@ def compare_preprod_artifact_size_analysis(
     project_id: int,
     org_id: int,
     artifact_id: int,
+    triggered_at: str | None = None,
     **kwargs: Any,
 ) -> None:
     logger.info(
@@ -266,7 +268,11 @@ def compare_preprod_artifact_size_analysis(
             except (ValueError, AttributeError, TypeError):
                 artifact_type_name = "unknown"
 
-            e2e_size_analysis_compare_duration = timezone.now() - artifact.date_added
+            # TODO: Remove artifact.date_added fallback once all producers pass triggered_at
+            start_time = (
+                datetime.fromisoformat(triggered_at) if triggered_at else artifact.date_added
+            )
+            e2e_size_analysis_compare_duration = timezone.now() - start_time
             metrics.distribution(
                 "preprod.size_analysis.compare.results_e2e",
                 e2e_size_analysis_compare_duration.total_seconds(),

src/sentry/preprod/tasks.py

@@ -679,6 +679,7 @@ def _assemble_preprod_artifact_size_analysis(
             "project_id": project.id,
             "org_id": org_id,
             "artifact_id": artifact_id,
+            "triggered_at": timezone.now().isoformat(),
         }
     )