Update docker.md readme to note memory requirements

[nvolungis]

Description

When attempting to get druid running inside Docker Desktop via the docker-compose file (both built from source and referencing the 0.17.0 published image), I experienced repeated crashing of the broker process with a 137 error code. 137 indicates an OOM error, so I notched up Docker's memory resource in 1gb increments (I believe the default is 2gb), landing on 6gb as a setting that works reliably.

As a newcomer with little knowledge of druid, it was quite difficult to identify exactly what the issue was initially. A note like this would have helped point me in the right direction.

---

This PR has:

• added documentation for new or modified features or behaviors.

[suneet-s]

Thanks for the pointer about error code 137! The integration tests need 4Gb to run.

What size cluster were you trying to start that it needed 6?

[nvolungis]

@suneet-s I was able to run just the broker consistently with 4gb. The whole cluster also runs with 4, but crashed occasionally. 6 seems to be the floor for consistency, at least on my machine. Perhaps noting a range of 4-6 would be more appropriate so people don't overshoot if unnecessary.

I'm running the cluster described in the docker-compose file unmodified which includes postgres, zookeeper, coordinator, broker, historical router, and middlemanager.

[clintropolis]

lgtm, thanks for the contribution! 👍