feat: add init sidecar for migrations and setup on compose up

Introduces a one-shot `init` service in docker-compose that runs Postgres
migrations, Neo4j index setup, and library-type seeding on every `up`.
Long-running services (`app`, `mcp`, `worker`) now depend on its
successful completion via `service_completed_successfully`, blocking the
stack on configuration errors (missing embedding model, dimension
mismatch, unreachable DB) rather than serving silent zero-result
searches.

Also standardizes reranker test fixtures to use the `/v1` OpenAI-style
base URL convention used across other service clients.

docs/PHASE_3_SEARCH_AND_RERANKING.md

@@ -61,6 +61,22 @@ POST http://pan.helu.ca:8400/v1/rerank
 }
 ```
 
+> **`LLMApi.base_url` convention.** Every Mnemosyne service client
+> (`EmbeddingClient`, `RerankerClient`, `vision.py`, `concepts.py`)
+> treats `base_url` as the **OpenAI-style `/v1` root** and appends a
+> path-only segment: `/embeddings`, `/rerank`, `/chat/completions`.
+> So a single `LLMApi` row with `base_url=http://pan.helu.ca:8400/v1`
+> serves both the embedding and the reranker endpoints — no per-purpose
+> duplication needed.
+>
+> Get this wrong (e.g. set `base_url=http://pan.helu.ca:8400` with no
+> `/v1`, or have a client prepend `/v1` locally) and you get a
+> double-prefixed URL like `…/v1/v1/rerank` that 404s silently —
+> `SearchService._rerank` catches the exception, the UI shows
+> "Re-rank: Skipped", and the search falls back to raw RRF order.
+> Check `results.reranker_skip_reason` on the search page for the
+> specific error.
+
 ## Deliverables
 
 ### 1. Search Service (`library/services/search.py`)

docs/mnemosyne.html

@@ -294,31 +294,37 @@ graph LR
             </div>
 
             <div class="alert alert-warning border-start border-4 border-warning">
-                <h4><i class="bi bi-lightning"></i> Neo4j Vector Indexes</h4>
-                <pre class="bg-light p-3 rounded mb-0"><code>// Chunk text+image embeddings (4096 dimensions, no pgvector limits!)
-CREATE VECTOR INDEX chunk_embedding FOR (c:Chunk)
+                <h4><i class="bi bi-lightning"></i> Neo4j Indexes (managed by <code>setup_neo4j_indexes</code>)</h4>
+                <p>Created by the <code>init</code> sidecar on every <code>docker compose up</code>. Vector dimensions come from the system embedding model's <code>vector_dimensions</code> field — the command fails if no model is configured. Current production model: <strong>Pan Synesis · qwen3-vl-embedding-2b · 2048d</strong>.</p>
+                <pre class="bg-light p-3 rounded mb-0"><code>// Chunk text+image embeddings (dimensions read from system embedding model)
+CREATE VECTOR INDEX chunk_embedding_index FOR (c:Chunk)
 ON (c.embedding) OPTIONS {indexConfig: {
-  `vector.dimensions`: 4096,
+  `vector.dimensions`: 2048,
   `vector.similarity_function`: 'cosine'
 }}
 
 // Concept embeddings for semantic concept search
-CREATE VECTOR INDEX concept_embedding FOR (con:Concept)
+CREATE VECTOR INDEX concept_embedding_index FOR (con:Concept)
 ON (con.embedding) OPTIONS {indexConfig: {
-  `vector.dimensions`: 4096,
+  `vector.dimensions`: 2048,
   `vector.similarity_function`: 'cosine'
 }}
 
 // Image multimodal embeddings
-CREATE VECTOR INDEX image_embedding FOR (ie:ImageEmbedding)
+CREATE VECTOR INDEX image_embedding_index FOR (ie:ImageEmbedding)
 ON (ie.embedding) OPTIONS {indexConfig: {
-  `vector.dimensions`: 4096,
+  `vector.dimensions`: 2048,
   `vector.similarity_function`: 'cosine'
 }}
 
-// Full-text index for keyword/BM25-style search
-CREATE FULLTEXT INDEX chunk_fulltext FOR (c:Chunk) ON EACH [c.text_preview]</code></pre>
+// Full-text indexes (BM25-style keyword search)
+CREATE FULLTEXT INDEX chunk_text_fulltext    FOR (c:Chunk)   ON EACH [c.text_preview]
+CREATE FULLTEXT INDEX concept_name_fulltext  FOR (c:Concept) ON EACH [c.name]
+CREATE FULLTEXT INDEX item_title_fulltext    FOR (i:Item)    ON EACH [i.title]
+CREATE FULLTEXT INDEX library_name_fulltext  FOR (l:Library) ON EACH [l.name]</code></pre>
+                <p class="mb-0 mt-3"><strong>Changing the embedding model or dimensions is a re-embedding event.</strong> Drop + recreate the vector indexes (<code>setup_neo4j_indexes --drop</code>) and re-queue all content for embedding. Old vectors at the previous dimension remain on the nodes until overwritten but are no longer indexed.</p>
             </div>
+
         </section>
 
         <!-- SECTION: CONTENT TYPES -->
@@ -521,10 +527,11 @@ flowchart TD
                         <div class="card-body">
                             <p>Cosine similarity via Neo4j vector index on Chunk and ImageEmbedding nodes.</p>
                             <pre class="bg-light p-2 rounded"><code>CALL db.index.vector.queryNodes(
-  'chunk_embedding', 30,
+  'chunk_embedding_index', 30,
   $query_vector
 ) YIELD node, score
 WHERE score > $threshold</code></pre>
+
                         </div>
                     </div>
                 </div>
@@ -548,9 +555,10 @@ RETURN c2, i2</code></pre>
                         <div class="card-body">
                             <p>Neo4j native full-text index for keyword matching (BM25-equivalent).</p>
                             <pre class="bg-light p-2 rounded"><code>CALL db.index.fulltext.queryNodes(
-  'chunk_fulltext',
+  'chunk_text_fulltext',
   $query_text
 ) YIELD node, score</code></pre>
+
                         </div>
                     </div>
                 </div>