[flagd] ContextEnrichment for inprocess improvements by extracting logic onto the channelmonitor

Currently our Context Enrichment is handled here:

https://github.com/open-feature/java-sdk-contrib/blob/122119bcbb4fe8e9bfe0de42feafaa54625c050f/providers/flagd/src/main/java/dev/openfeature/contrib/providers/flagd/resolver/process/storage/connector/grpc/GrpcStreamConnector.java#L107-L124

We always try to fetch the metadata before we reestablish a connection, during our stream listening logic.

This might be a little bit tricky, as we are always relying on the same process, and if eg. the timeline exceeds etc. the whole process breaks.

The question is how important it is to fetch this metadata, as we could theoretically also do this unary call when the channel state changes to READY, instead of always during the loop.

This would decouple our sync from the metadata, and might increase our robustness.

WDYT?

cc: @guidobrei @toddbaert @beeme1mr @alexandraoberaigner @chrfwow

Start in providers/flagd/src/main/java/dev/openfeature/contrib/providers/flagd/resolver/process/storage/connector/grpc/GrpcStreamConnector.java, especially the metadata handling around lines 107–124. Read the stream listening and channel state logic, then assess moving the unary metadata fetch to the channel's READY transition. Done means synchronization no longer depends on fetching metadata inside the stream loop and the reconnection behavior remains robust.