Lot's of CPU usage when re-sending many nacked events

[t-persson]

Description

When resending nacked events the Eiffellib will iterate over all unsent and send them in a busy-loop. This causes the machine running the Eiffellib to use 100% CPU when re-sending many.

Motivation

We're seeing that some events are not sent due to missing a health-check from RabbitMQ. When debugging this I noticed that our machines CPUs are being hogged by the re-send.

Exemplification

We would not be using 100% CPU unnecessarily

Benefits

Since we are seeing events that are not being sent it would be a benefit to actually have them being sent

Possible Drawbacks

Adding a sleep in the re-send loop would make it slower, but adding a low enough sleep that it does not hog the CPU and is still quick should be possible