Ingest rows with Snowflake's high-performance Snowpipe Streaming SDK using channels

Steps

1. Configure OpenClient with DB, SCHEMA, and PIPE — unlike the classic SDK, a PIPE object is required and central to the high-performance architecture.
2. Open a channel with openChannel(channelName, offsetToken) — only the channel name is required, avoiding the extra round trip the classic SDK needed to fetch the last committed offset.
3. Append rows with appendRow(row, offsetToken) or appendRows(rows, startOffsetToken, endOffsetToken) — note the method is renamed from insertRows and no longer returns an AppendResult.
4. Track progress asynchronously with getChannelStatus(), and control when buffered data lands using waitForFlush(), waitForCommit(), and initiateFlush() instead of polling per-insert results.
5. Send strictly typed values for ARRAY and VARIANT columns — the high-performance SDK type-checks them rather than auto-parsing JSON-as-string the way the classic SDK tolerated.