Today counted_keyword field type support synthetic source via the fallback mechanism. This means we support it via the ignore source infrastructure, and upon index we additionally store an ignore source stored field. This isn't needed because counted keyword also stores the content in doc values, and we can synthesize from that as well. So storing it also in ignored source is kind of wasteful, because we would store these kind of fields twice.
We should just support synthetic source for counted_keyword field type natively.
Today counted_keyword field type support synthetic source via the fallback mechanism. This means we support it via the ignore source infrastructure, and upon index we additionally store an ignore source stored field. This isn't needed because counted keyword also stores the content in doc values, and we can synthesize from that as well. So storing it also in ignored source is kind of wasteful, because we would store these kind of fields twice.
We should just support synthetic source for counted_keyword field type natively.