fix(embedder): use correct sentence-transformers method name

[l984-451]

Problem

On fresh startup, the embedder fails to initialize with:

[Embedder] Failed to load model: 'SentenceTransformer' object has no attribute 'get_embedding_dimension'
[AnalysisWorker] Embedder unavailable: 'SentenceTransformer' object has no attribute 'get_embedding_dimension'

This silently disables semantic search across the timeline (FTS5 keyword search still works, but embedding-backed search via engine/embedder.py:embed_activity never runs).

Fix

engine/embedder.py:31 calls self._model.get_embedding_dimension() but sentence_transformers.SentenceTransformer (versions 3.x, including the pinned 3.3.1) exposes the method as get_sentence_embedding_dimension(). One-character… well, fifteen-character fix.

Verified locally with sentence-transformers==3.3.1:

>>> from sentence_transformers import SentenceTransformer
>>> m = SentenceTransformer('all-MiniLM-L6-v2')
>>> hasattr(m, 'get_embedding_dimension')
False
>>> hasattr(m, 'get_sentence_embedding_dimension')
True
>>> m.get_sentence_embedding_dimension()
384

After the fix, embedder boots cleanly and status.analysis.embedder_available is true.