[Inference] The semantic_text type index options parameters

[weizijun]

The semantic_text type is very useful for the AI Search.
When inference_id is text_embedding, the semantic_text internal field is dense_vector type.
But now the semantic_text can't support index_options of the dense_vector type.
User can't set quantification type or m, ef parameters.
So I added the index_options in the semantic_text type.

[elasticsearchmachine]

Pinging @elastic/search-eng (Team:SearchOrg)

[elasticsearchmachine]

Pinging @elastic/search-relevance (Team:Search - Relevance)

[jimczi]

Thanks, @weizijun, and apologies for the delayed response.

We're actively working on adding index options to the semantic text field in [this PR](#119967). During this process, we realized that validating these index options requires an underlying infrastructure change, which we're addressing in [this PR](#121106). This additional work has caused some delays.

Once the infrastructure change is merged, we'll update the original PR to incorporate the necessary logic, ensuring that invalid index options are detected and rejected early.

[weizijun]

Thanks, @weizijun, and apologies for the delayed response.

We're actively working on adding index options to the semantic text field in [this PR](#119967). During this process, we realized that validating these index options requires an underlying infrastructure change, which we're addressing in [this PR](#121106). This additional work has caused some delays.

Once the infrastructure change is merged, we'll update the original PR to incorporate the necessary logic, ensuring that invalid index options are detected and rejected early.

Thanks @jimczi , what a coincidence, this feature is really useful and I hope it can be merged soon.

[kderusso]

Hi @weizijun , thanks for the work on this PR. I'm going to close it, as we've merged a separate feature that will address this need in an upcoming release: #119967

This will allow you to set index_options for dense vector fields.

[weizijun]

Hi @weizijun , thanks for the work on this PR. I'm going to close it, as we've merged a separate feature that will address this need in an upcoming release: #119967

Thanks @kderusso !