docs: remove unsubstantiated SOTA claim from README (#3)

* fix: Use filename-based GitHub Actions badge URL

The workflow-name-based badge URL was showing "no status" because
GitHub requires workflow runs on the specified branch. Using the
filename-based URL format (actions/workflows/publish.yml/badge.svg) is
more reliable and works regardless of when the workflow last ran.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

* docs: remove unsubstantiated SOTA claim from README

Changed "state-of-the-art VLM" to just "Qwen3-VL model" since SOTA
claims require citations or benchmark comparisons to be defensible.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>

Author: abrichr

README.md

@@ -14,7 +14,7 @@ Multimodal demo retrieval using VLM embeddings for GUI automation.
 
 **Key Features:**
 - **Multimodal Embeddings**: Embed text, images, or both into a shared vector space
-- **Qwen3-VL-Embedding Support**: Primary embedder using Alibaba's state-of-the-art VLM
+- **Qwen3-VL-Embedding Support**: Primary embedder using Alibaba's Qwen3-VL model
 - **Matryoshka Representation Learning (MRL)**: Flexible embedding dimensions (512-8192)
 - **FAISS Integration**: Fast similarity search with support for large demo libraries
 - **Persistence**: Save and load indices with embeddings and metadata