[FEATURE]Create MV index target using existing opensearch index

[YANG-DB]

Is your feature request related to a problem? There are some cases which existing indices or index templates are needed to be used as the ingestion target sync of the MV query.

For example:

• existing index template contains specific data-types supported only in opensearch
• existing index contains data that needs to be appended with the MV query results
• existing index has former version of the MV and need to continue holding the data with the updated MV

What solution would you like? The MV creation metadata specification would have the next fields: using_existing_index = opensearch_index_name

WITH (
  auto_refresh = true,
  refresh_interval = '15 Minute',
  checkpoint_location = '{s3_checkpoint_location}',
  watermark_delay = '1 Minute',
  using_existing_index = 'opensearch_index_name',
  extra_options = '{ "{table_name}": { "maxFilesPerTrigger": "10" }}'
)

For the index name to be compliant with Flint's naming convention - the MV name would be added to the index as an index alias

Do you have any additional context?

• https://opensearch.org/docs/latest/im-plugin/index-alias/
• https://github.com/opensearch-project/opensearch-spark/blob/main/docs/index.md#create-index-options

• ---

[noCharger]

Although not explicitly stated in https://github.com/opensearch-project/opensearch-spark/blob/main/docs/index.md#flint-index-specification, the CREATE MATERIALIZED VIEW command does not indicate that any existing materialized view can be utilized. Alter the MV definition in with clause seems to be acheivable.

[penghuo]

Try to understand the use case of create MV with indexing index. and what is expected workflow?

existing index template contains specific data-types supported only in opensearch

We can extend Spark data type to support OpenSearch data type.

existing index contains data that needs to be appended with the MV query results

UNION can solve this problem.