Big value serialization - Add Flushing Support While Iterating Bucket Entries

[adiholden]

Currently, we only flush during serialization of container data types (e.g., sets, hashes, etc.).
We intentionally avoid flushing in the middle of a string. However, if a bucket contains only string entries, we end up not flushing at all during that bucket's serialization. This can lead to exceeding the serialization_max_chunk_size.

To address this, we should introduce flushing while iterating over individual entries within a bucket—even for simple types like strings.