fix: ensure correct handling of mock endpoints

[HalosGhost]

This ensures that, regardless of what has been specified for mocking, any module with no endpoints defined in the config will have a server started for it in a mock system test.

[HalosGhost]

awaiting fix for test breakage in 211bbc9 to rebase and rerun CI.

[AlexRamRam]

utACK

The intended changes look solid, as intended.

A comment that may be beyond the scope of this PR:
It may be valuable if the mock_system provides some level of self-integrity by ensuring that the input configuration is at least theoretically workable for one of the existing architectures. This would require some documentation and checks with error messages. I may be missing some context, but the self-integrity feature may make this module even more compelling, especially as a c++ module as opposed to a shell utility aimed to serve a similar purpose.

[HalosGhost]

@AlexRamRam note that the mock_system is a testing utility which allows us to mock N components while running one or more “real” components so we can do integration/boundary tests.

However, a utility that parses a config and determines if it is even statically-valid for the specified architecture would be an excellent tool to have. (Though, it probably requires us to finalize an answer to #218 before we could reasonably implement one.)

[maurermi]

T-ACK. Tests run as expected, and this improves our test suite by allowing endpoints to be mocked properly. Left one comment about the changes to .clang-tidy to consider but think this is good to merge.