Query1_optimized.sql

start t=node:types(id={id})
match
      (t)<-[:has_type]-(u)<-[:metric_unit]-(ma)<-[:metric_activity]-(m)<-[cm:cycle_metric]-(c)
where
      c.measureDate>={measureStartDate} and c.measureDate <= {measureEndDate}
with distinct m, ma, u, c
match
      (u)<-[:alert_for_unit]-(a)
where
      a.alertDate=c.measureDate and a.fromEntityType={type}
with a,ma,u
match
      (i)-[:has_result]->(ir)<-[:for_inspection_result]-(a)
where
      (i)<-[:metric_inspection]-(ma)
      and (a)-[:alert_for_inspection]->(i)
return ir.value as alertValue, count(ir.value) as alertCount, u.id as unitId, u.name as unitName

The distinct in line 6 kills the query lazyness and also performance, how does it perform if you leave that off?

It does perform better, but I do believe I need that distinct- it is worth a second look though to confirm that.