[Tasks] disable chat completion automatic specs generation (#1635)

#1610 (comment)

let's not generate the chat completion specs from TGI, the schema is
currently stable, and we want to ensure full OpenAI-compatibility for
chat completion, and we ve already updated the schema manually, so
relying on automatic generation is unnecessary now.

Author: hanouticelina

.github/workflows/update-specs.yml

@@ -47,7 +47,6 @@ jobs:
           body: |
             This PR updates the @huggingface/tasks specs. It has been generated by running:
             ```sh
-            pnpm run inference-tgi-import
             pnpm run inference-tei-import
             pnpm run inference-codegen
             ```