Speed up stopping containers

[lognaturel]

Closes #416

What has been done to verify that this works as intended?

Did a docker compose build docker compose up -d docker compose stop after each commit to make sure each was minimal and effective. It's still on there if someone wants to do a sanity check.

Why is this the best possible solution? Were any other approaches considered?

The underlying issue is that signals weren't being proxied to the intended processes. More at https://stackoverflow.com/a/32261019/137744. This was addressed by running all the various processes with exec.

Additionally, the Enketo dockerfile needed to be updated to use the JSON-array syntax so it wouldn't be called with /bin/sh -c which is immune to SIGTERM

How does this change affect users? Describe intentional changes to behavior and behavior that could have accidentally been affected by code changes. In other words, what are the regression risks?

This is an ops change. Worst case would be things don't stop or come up. I think if it works on dev it's likely to work anywhere.

I'm not entirely sure about risk beyond things not stopping or coming up. I keep seeing things about the PID 1 zombie reaping problem but I can't tell if this changes anything with respect to that. This post suggests that it doesn't change anything. This tini post has more about zombie processes. Given my understanding of the processes we launch, I believe it's unlikely that any secondary processes likely to become zombies are being launched.

Does this change require updates to documentation? If so, please file an issue here and include the link below.

No.

Before submitting this PR, please make sure you have:

• verified that any code or assets from external sources are properly credited in comments or that everything is internally sourced

[sadiqkhoja]

postgres14 container is taking 10s for me. I think SIGTERM is not being passed to the postgres process

[lognaturel]

Good catch about the db container. Let me look at that separately because it doesn't affect the Cloud setup and I'd like to make sure I understand that process specifically.

Are you feeling pretty confident that this is low-risk? Like I wrote in the risks section, I am not entirely sure that there aren't possible side effects I haven't considered.

[sadiqkhoja]

I don't see any risk of zombie processes here as we are not spawning processing on the runtime.

[lognaturel]

Have talked it through with @yanokwa and am back to feeling like it's low-risk. Moreover, I don't think we have a great way of identifying possible issues other than putting it out there on different systems.

[lognaturel]

Oh great, thanks @sadiqkhoja.