Don't try to catch up after slowdown when using fixed rate

With fixed rate limit, if there is sudden slowdown and the
database can't keep up, latte should not try to issue
more requests in burst after the slow-down is over.
With this fix it will keep the maximum set rate limit.

(cherry picked from commit 1b252a3)

src/exec.rs

@@ -12,6 +12,7 @@ use std::pin::Pin;
 use std::sync::Arc;
 use std::task::{Context, Poll};
 use std::time::Instant;
+use tokio::time::MissedTickBehavior;
 use tokio_stream::wrappers::IntervalStream;
 
 use crate::chunks::ChunksExt;
@@ -22,8 +23,10 @@ use crate::{
 
 /// Returns a stream emitting `rate` events per second.
 fn interval_stream(rate: f64) -> IntervalStream {
-    let interval = tokio::time::Duration::from_nanos(max(1, (1000000000.0 / rate) as u64));
-    IntervalStream::new(tokio::time::interval(interval))
+    let period = tokio::time::Duration::from_nanos(max(1, (1000000000.0 / rate) as u64));
+    let mut interval = tokio::time::interval(period);
+    interval.set_missed_tick_behavior(MissedTickBehavior::Skip);
+    IntervalStream::new(interval)
 }
 
 /// Runs a stream of workload cycles till completion in the context of the current task.