docs: llama.cpp across nodes, spice chat via /v1/responses, embeddings after filters, Glue Iceberg credential refresh (#2455) * docs: llama.cpp across nodes, spice chat via /v1/responses, embeddings after filters, Glue Iceberg credential refresh Documents spiceai/spiceai#15584, #15673, #15651 and #15667 in the vNext docs. Co-Authored-By: Claude <noreply@anthropic.com> * docs: replace nonexistent spice chat --responses flag on 2.4.x and 2.3.x web-search pages spice chat in v2.4.0 and v2.3.4 has no --responses flag and sends every message to /v1/chat/completions. Co-Authored-By: Claude <noreply@anthropic.com> * docs(azure): use the unprefixed openai_responses_tools key for Azure OpenAI models The runtime reads only the literal openai_responses_tools key on the Responses path (vNext, v2.4.0, v2.3.4); azure_openai_responses_tools is never read. The Azure example also sets tools: auto, which hosted tools require. Co-Authored-By: Claude <noreply@anthropic.com> * docs: qualify Glue Iceberg credential refresh for directly supplied tokens; version-relative spiced link --------- Co-authored-by: Emma Spice <emma@EmmasMacStudio.localdomain> Co-authored-by: Claude <noreply@anthropic.com>