fix: killing subprocess only if prism is running in multiprocess mode

[karol-maciaszek]

Closes #644

Checklist

• Tests have been added (for bug fixes / features)
• Docs have been added / updated (for bug fixes / features)

What kind of change does this PR introduce?

Bug fix

What is the current behavior? What is the new behavior?

Cli is trying to kill subprocess even if it is not running in multi-process mode. It results in no nice-looking line ✖ fatal Cannot read property 'kill' of undefined.

This fix moves logging of errors to the place where the context (single/multiprocess) is known. .kill will be invoked only in the latter context.

Does this PR introduce a breaking change?

No

[lag-of-death]

it's OK when the changelog is adjusted 👍

[lag-of-death]

hm.. just curious if we could do sth like:

====server====
mock -p 4010
====command====
mock -p 4010

in an e2e test? maybe the change could be tested like this? WDYT @StefanDywersant ?

[XVincentX]

Minor comments, good to go. I'll leave the merge to you.

[karol-maciaszek]

hm.. just curious if we could do sth like:

====server====
mock -p 4010
====command====
mock -p 4010

in an e2e test? maybe the change could be tested like this? WDYT @StefanDywersant ?

I doubt whether I can match log lines here. I also can't start two prism instances here in order to reproduce the error.