Function index_dataflow 

pub fn index_dataflow(
    source_id: GlobalId,
    index_id: GlobalId,
    shard: ShardId,
    location: PersistLocation,
    desc: RelationDesc,
    key_cols: Vec<usize>,
    as_of: Timestamp,
    shard_upper: Timestamp,
) -> Result<DataflowDescription<RenderPlan, CollectionMetadata>>

Build a single-index dataflow over a persist shard.

Thin sugar over DataflowBuilder for the common shape: import the collection backed by shard as source_id, set as_of, and export an index index_id arranging the collection by key_cols.

shard_upper is the exclusive upper bound of the shard’s written data; see PersistSource::upper.