refactor: startup config validation

[lvlcn-t]

Motivation

Currently the target manager config has a Validate method that is never consumed.

This PR is part of #66 which will implement another target manager type and therefore needs more target manager config validation. That's also when I noticed that the target manager config is never validated. 😄

Changes

I've refactored the startup config validation so every startup config component has its own Validate() method.

• fix: validate target manager config
• refactor: split config validation in smaller methods
• chore: use errors instead of ok booleans

For additional information look at the commits.

Tests done

I've adjusted the existing unit tests.

• Unit tests succeeded
• E2E tests succeeded

TODO

• I've assigned this PR to myself
• I've labeled this PR correctly

[lvlcn-t]

The actions are failing because the PR comes from my fork.

All checks are succeeding here, where I've implemented the whole issue of #66 (also with the changes of this PR): https://github.com/caas-team/sparrow/tree/feat/git-tarman

[puffitos]

LGTM. I don't appreciate the log in the validate function, that's only code we'll forget to update when the logic changes. We should just return a (structured) error. The standardized error text is OK for now.

Thanks for adding the godocs as well and refactoring the tests :)

[lvlcn-t]

@puffitos I agree with you, maybe we should change that in the future to returning to a structured error like this:

// ErrInvalidConfig is an error for invalid configuration
type ErrInvalidConfig struct {
	Field string
	Err   error
}

// Error returns the error message
func (e ErrInvalidConfig) Error() string {
	return fmt.Sprintf("invalid config for %q: %v", e.Field, e.Err)
}

But since it makes no difference for the user, I'll implement this in another PR in the near future.
(I want to merge this to continue opening PR's for #66 😄 )

[puffitos]

@lvlcn-t the struct seems appropriate. The #111 will take priority in this iteration, just FIY.