Tag as text-classification (pipeline_tag only)

Carries over the one change from #8 that is not already on `main`, as a single line on current `main`: `pipeline_tag: schema-extraction` becomes `text-classification`. Everything else in #8 (the `library_name`, the `decision-model` tag, the banner links) is left as `main` has it, because those were changed deliberately in #10, #11 and the Sep 28 banner commits.

**Why:** when this card was uploaded, the Hub's metadata validator warned that `schema-extraction` is not in its list of official pipeline tags. Hub task filters and search key off that list, and this checkpoint is used for classification (`classify_text`), which `text-classification` lists correctly.

**Trade-off, for the model owners to decide:** #11 chose `schema-extraction` on purpose, to match the native Transformers `pipeline("schema-extraction", ...)` class (`Gliner2ForSchemaExtraction`). This PR does not touch `config.json`, so that loading path is unaffected, but the card tag and that pipeline name would no longer be identical.

@MattThomas-fastino could you decide whether to take this? Credit for the original suggestion goes to @bkinge (#8).

README.md

@@ -3,7 +3,7 @@ library_name: transformers
 license: apache-2.0
 language:
 - en
-pipeline_tag: schema-extraction
+pipeline_tag: text-classification
 tags:
 - gliner2
 - Text classification