Clarified the word 'latency' w.r.t MPMC queues

Author: p-ranav

README.md

@@ -117,7 +117,7 @@ See [benchmarks](https://github.com/p-ranav/binary_log/blob/master/README.md#ben
   - There are numerous well-known lock-free queues available for this purpose ([moody::concurrentqueue](https://github.com/cameron314/concurrentqueue), [atomic_queue](https://github.com/max0x7ba/atomic_queue) etc.) - let the user choose the technology they want to use.
   - The latency of enqueuing into a lock-free queue is large enough to matter
     - Users who do not care about multi-threaded scenarios should not suffer the cost
-    - Looking at the [atomic_queue benchmarks](https://max0x7ba.github.io/atomic_queue/html/benchmarks.html), the average latency across many state-of-the-art multi-producer, multi-consumer queues is around 150-250 ns
+    - Looking at the [atomic_queue benchmarks](https://max0x7ba.github.io/atomic_queue/html/benchmarks.html), the average round-trip latency across many state-of-the-art multi-producer, multi-consumer queues, to send and receive a 4-byte integer (between 2 threads, using 2 queues) is around 150-250 ns.
 * Avoid writing static information more than once
   - Examples of static information: the format string, the number of format args, and type of each format arg
   - Store the static information in an "index" file