fix: ignore schedule missed-run counters in workspace compare (#11570)

Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

backend/windmill-api-integration-tests/tests/workspace_comparison.rs

@@ -756,6 +756,13 @@ async fn test_compare_workspaces_trigger_and_schedule(db: Pool<Postgres>) -> any
     )
     .execute(&db)
     .await?;
+    // The scheduler's missed-occurrence counters only ever move on the enabled side.
+    sqlx::query(
+        "UPDATE schedule SET late_run_streak = 1, missed_occurrences = 2, last_missed_at = NOW()
+         WHERE workspace_id = 'test-workspace' AND path = 'f/sch/runtime_only'",
+    )
+    .execute(&db)
+    .await?;
 
     // ------ Schedule: config change (script_path) in fork. Should diff.
     sqlx::query!(
@@ -2537,7 +2544,10 @@ async fn test_rename_records_the_vacated_path(db: Pool<Postgres>) -> anyhow::Res
             .await?;
         let status = resp.status();
         let hash = resp.text().await?;
-        assert!(status.is_success(), "script deploy failed: {status} — {hash}");
+        assert!(
+            status.is_success(),
+            "script deploy failed: {status} — {hash}"
+        );
         parent_hash = Some(hash.trim().trim_matches('"').to_string());
     }
     let mut vacated = None;

backend/windmill-api-workspaces/src/workspaces.rs

@@ -14050,6 +14050,12 @@ const TRIGGER_COMPARE_IGNORE: &[&str] = &[
     "last_server_ping",
     "extra_perms",
     "permissioned_as",
+    // Schedule counters the scheduler bumps as occurrences are missed. A fork's
+    // schedules are disabled so theirs stay at the defaults, and no deploy carries
+    // them across: compared, a parent schedule that ever ran late differs forever.
+    "late_run_streak",
+    "missed_occurrences",
+    "last_missed_at",
     // Server-managed fields that the merge feature treats as workspace-local
     // (regenerated by the deploy handler): GCP `subscription_id` is rewritten
     // to `windmill_<workspace_id>_<path>` in `CreateUpdate` mode, and Azure's