Refinements on the semantic_text field type content

This is a follow-up issue for https://github.com/elastic/docs-content/issues/3836.

In that issue, we restructured the content for the semantic_text field type. A few additional refinements are still needed, listed below.

**To validate these refinements ideas, reach out to SMEs first.**

### Create a dedicated place for limitations and troubleshooting

Create a new subpage (for example, Limitations and troubleshooting) and move or link the following content there:

- Limitations (currently on the Reference page)
- Document count discrepancy in _cat/indices (currently on the Reference page)
- Retrieve semantic_text field embeddings: keep this content in its current location, but add a brief explanation that it is intended for debugging, and link to it from the new page.

### Rework the Updates and partial updates section into a how-to guide

This section is currently on the Ingestion page. Restructure it to be more like a how-to guide, including clear explanations and code examples.

### Clarify the distinction between preconfigured and default endpoints

As @maxjakob suggested in [this comment](https://github.com/elastic/elasticsearch/pull/138571#discussion_r2609918943), add the following information to the [Configure inference endpoints](https://www.elastic.co/docs/reference/elasticsearch/mapping-reference/semantic-text-setup-configuration#configure-inference-endpoints) section to clarify the distinction between preconfigured and default endpoints:

- Preconfigured endpoints are automatically created by Elasticsearch on startup. These are many in an ES cluster. Examples: .elser-2-elasticsearch, .elser-2-elastic or .jina-embeddings-v3 (about to be released).
- Some features like semantic_text have default endpoints configured. There is always just one per features. Examples: semantic_text uses .elser-2-elasticsearch, text_similarity_reranker will use .jina-reranker-v2 (soon).