scexec: don't set running_status on GC job during init

The running_status is only supposed to be set while the job is running,
so it was a mistake to set it when creating the job.

Not setting it also deflakes a test which relies on the running_status
to know if the GC has begun (and deleting the data has completed).

Release note: None

pkg/sql/drop_test.go

@@ -417,6 +417,14 @@ func TestDropIndex(t *testing.T) {
 	s, sqlDB, kvDB := serverutils.StartServer(t, params)
 	defer s.Stopper().Stop(context.Background())
 
+	_, err := sqlDB.Exec(`SET CLUSTER SETTING sql.gc_job.wait_for_gc.interval = '1s';`)
+	require.NoError(t, err)
+
+	// Refresh protected timestamp cache immediately to make MVCC GC queue to
+	// process GC immediately.
+	_, err = sqlDB.Exec(`SET CLUSTER SETTING kv.protectedts.poll_interval = '1s';`)
+	require.NoError(t, err)
+
 	// Disable strict GC TTL enforcement because we're going to shove a zero-value
 	// TTL into the system with AddImmediateGCZoneConfig.
 	defer sqltestutils.DisableGCTTLStrictEnforcement(t, sqlDB)()
@@ -426,6 +434,9 @@ func TestDropIndex(t *testing.T) {
 		t.Fatal(err)
 	}
 	tableDesc := desctestutils.TestingGetPublicTableDescriptor(kvDB, keys.SystemSQLCodec, "t", "kv")
+	if _, err := sqltestutils.AddImmediateGCZoneConfig(sqlDB, tableDesc.GetID()); err != nil {
+		t.Fatal(err)
+	}
 	tests.CheckKeyCount(t, kvDB, tableDesc.TableSpan(keys.SystemSQLCodec), 3*numRows)
 	idx, err := catalog.MustFindIndexByName(tableDesc, "foo")
 	if err != nil {

pkg/sql/schemachanger/scexec/gc_jobs.go

@@ -233,15 +233,13 @@ func createGCJobRecord(
 			descriptorIDs = append(descriptorIDs, details.ParentID)
 		}
 	}
-	runningStatus := jobs.RunningStatus("waiting for MVCC GC")
 	return jobs.Record{
 		JobID:         id,
 		Description:   "GC for " + description,
 		Username:      userName,
 		DescriptorIDs: descriptorIDs,
 		Details:       details,
 		Progress:      jobspb.SchemaChangeGCProgress{},
-		RunningStatus: runningStatus,
 		NonCancelable: true,
 	}
 }