Fix schema drift: fresh installs were broken

- Create heat_cycles table in init.js (breeding.js used it but nothing created it)
- Add litters columns the routes actually use (breeding_date, whelping_date, puppy_count)
- Add health_records columns used by puppy logs (record_date, description)
- Puppy log insert now also sets NOT NULL test_date
- Migration 002 no-ops on a fresh DB instead of crashing on missing dogs table
- init.js honors DB_PATH and creates the data dir, so migrations and the app
  always target the same file

Verified: full heat-cycle/litter/puppy-log CRUD passes against a virgin database.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

server/db/migrations.js

@@ -52,6 +52,13 @@ class MigrationRunner {
     return columns.some(col => col.name === 'sire' || col.name === 'dam');
   }
 
+  // Check if a table exists (fresh DBs have none — migrations must no-op)
+  tableExists(name) {
+    return !!this.db.prepare(
+      "SELECT name FROM sqlite_master WHERE type='table' AND name = ?"
+    ).get(name);
+  }
+
   // Check if litter_id column exists
   hasLitterIdColumn() {
     const columns = this.db.prepare("PRAGMA table_info(dogs)").all();
@@ -187,7 +194,12 @@ class MigrationRunner {
   // Migration 2: Add litter_id column if missing
   migration002_addLitterIdColumn() {
     console.log('[Migration 002] Checking for litter_id column...');
-    
+
+    if (!this.tableExists('dogs')) {
+      console.log('[Migration 002] Fresh database (no dogs table yet), skipping — init will create it');
+      return;
+    }
+
     if (this.hasLitterIdColumn()) {
       console.log('[Migration 002] litter_id column already exists, skipping');
       return;