ticdc: clarify descriptions of index-value and columns (#21616)

Author: hfxsd

ticdc/ticdc-sink-to-kafka.md

@@ -280,8 +280,8 @@ For example, for a dispatcher like `matcher = ['test.*'], topic = {schema}_{tabl
 You can use `partition = "xxx"` to specify a partition dispatcher. It supports five dispatchers: `default`, `index-value`, `columns`, `table`, and `ts`. The dispatcher rules are as follows:
 
 - `default`: uses the `table` dispatcher rule by default. It calculates the partition number using the schema name and table name, ensuring data from a table is sent to the same partition. As a result, the data from a single table only exists in one partition and is guaranteed to be ordered. However, this dispatcher rule limits the send throughput, and the consumption speed cannot be improved by adding consumers.
-- `index-value`: calculates the partition number using either the primary key, a unique index, or an index explicitly specified by `index`, distributing table data across multiple partitions. The data from a single table is sent to multiple partitions, and the data in each partition is ordered. You can improve the consumption speed by adding consumers.
-- `columns`: calculates the partition number using the values of explicitly specified columns, distributing table data across multiple partitions. The data from a single table is sent to multiple partitions, and the data in each partition is ordered. You can improve the consumption speed by adding consumers.
+- `index-value`: calculates the partition number using either the primary key, a unique index, or an index explicitly specified by `index`, distributing table data across multiple partitions. The data from a single table is sent to multiple partitions, and the data in each partition is ordered. You can improve the consumption speed by adding consumers. This dispatcher also ensures that updates to the same row are sent to the same partition, which guarantees ordered processing for that row.
+- `columns`: calculates the partition number using the values of explicitly specified columns, distributing table data across multiple partitions. The data from a single table is sent to multiple partitions, and the data in each partition is ordered. You can improve the consumption speed by adding consumers. This dispatcher also ensures that updates to the same row are sent to the same partition, which guarantees ordered processing for that row.
 - `table`: calculates the partition number using the schema name and table name.
 - `ts`: calculates the partition number using the commitTs of the row change, distributing table data across multiple partitions. The data from a single table is sent to multiple partitions, and the data in each partition is ordered. You can improve the consumption speed by adding consumers. However, multiple changes of a data item might be sent to different partitions and the consumer progress of different consumers might be different, which might cause data inconsistency. Therefore, the consumer needs to sort the data from multiple partitions by commitTs before consuming.