We can't detect why a provider function is missing is all cases, so
get rid of the extra logic trying to do that.
We also admit here that we need to allow a provider function within a
variable validation condition, because that is not the only way these
could fail during init, and it's otherwise valid during the next phase.
We can't fully detect why a module source or version might be unknown,
so instead use a simpler error that covers all possible reasons. If we
do find a way to get more specific in the future, we can extent it then,
but the current errors were trying to be specific when they couldn't.
* refactor: Split temp file creation logic out of FakeInstallablePackageMeta & FakePackageMetaViaHTTP. Handle cleanup internally.
* refactor: Use 'Must' helpers in `newMockProviderSourceUsingTestHttpServer`
* refactor: Update tests using `newMockProviderSource` to not handle 'close' callback
We previously used a loader -> BuildConfig flow to load configuration.
This commit changes most (but not all yet) flows to use the new
graph-based approach. Instead of simply recursively loading the modules,
we now need to take a stepped approach:
1. Load the root module
2. Collect the variables and their values
3. Build the configuration with the graph-based approach
Because this approach relies on different parts from different packages,
it can't easliy be done within the `configload` package. So, now we do
most of in the backend or command.
