Try harder to get the microgrid metadata #51
Description
What's needed?
Right now we are just calling the API and if the call fails, for any reason, we just use an empty metadata and move on. We need to make sure we have microgrid metadata if we can get any, even if the fetching is delayed.
Proposed solution
We should differentiate between permanent errors (like invalid argument, not implemented, etc.) from transient errors (like unavailable, failed precondition, etc.), and if we got a transient error, we should probably retry a few times.
Use cases
No response
Alternatives and workarounds
No response
Additional context
Another approach would be that the client itself somehow caches the metadata, and have a background task to get it if the first try failed. For this the client should probably have a much higher level, probably focusing in one particular microgrid, and caching the components and connections too.