fix: fix Github workflow caches to restore successfully more often

[ansgarm]

This PR

• Stops using random temporary directories for caches, as these make cache hits impossible (the cached paths are part of the "cache version" which needs to match in order for caches to be restored)
• Starts using the same cache directories for caches of the same type (e.g. /usr/local/share/.cache/go for the Go cache). This allows us to get cache hits between different jobs as a fallback
• Starts using distinct cache directories for all caches of the same type (types being terraform, go, yarn and providerSchema).
• Re-names the cache keys to allow cache hits using a hierarchical restore-keys definition
• Defines restore-keys for caches of the same type. This allows us to restore e.g. a yarn cache from a different job if there's no cache already for the current job (while the cache might not match 100%, it will still improve the results)
• fixes the specified Terraform version for provider integration tests (used to access a matrix variable that does not exist for these tests; probably a copy&paste error)
• removes the usage of YARN_CACHE_FOLDER which is not necessary, since our cache is in the yarn cache dir already
• removes GOCACHE for jobs that don't require it

[mutahhir]

Approving, with comments regarding improvements

[github-actions]

I'm going to lock this pull request because it has been closed for 30 days. This helps our maintainers find and focus on the active issues. If you've found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.