fix: Lazy grammar loading and quantized embeddings to prevent V8 WASM OOM

Fixes #54 — `codegraph init -i` crashes with "Fatal process out of memory: Zone"
on large codebases because all 16 tree-sitter WASM grammar modules were compiled
upfront by V8, exhausting the WASM Zone allocator.

Changes:
- initGrammars() now only initializes the tree-sitter WASM runtime (Parser.init()),
  no longer eagerly loads all grammar files
- New loadGrammarsForLanguages() loads only grammars for languages actually present
  in the project (e.g. a Dart project loads ~2-3 grammars instead of 16)
- Orchestrator detects needed languages after file scan, before parsing begins
- Embedding pipeline now uses quantized model (~67MB vs ~270MB) to further reduce
  WASM memory pressure when embeddings are enabled

src/vectors/embedder.ts

@@ -127,8 +127,10 @@ export class TextEmbedder {
       env.allowRemoteModels = false;
     }
 
-    // Load the pipeline
+    // Load the pipeline with quantized model to reduce WASM memory pressure.
+    // Quantized (int8/uint8) is ~4x smaller than FP32 with minimal quality loss.
     this.pipeline = await pipeline('feature-extraction', this.modelId, {
+      quantized: true,
       progress_callback: this.showProgress
         ? (progress: { status: string; file?: string; progress?: number }) => {
             if (progress.status === 'progress' && progress.file && progress.progress) {