Prune ghost words from affix expansion and stop frequency loader creating words

- Add PruneOrphanWords post-import step that deletes words with no
  definitions, synonyms, translations, pronunciations, etymology,
  synsets, or frequency data
- Remove INSERT path from CETEMPublico loader — frequency data should
  annotate existing words, not create new ones (was inflating pt-PT
  from 136K real words to 718K)
- Add hasLetter filter to hunspell, affix, and dicionario loaders to
  block punctuation entries like ".", ",", "(" from entering the DB
- Move hasLetter helper to scowl.go alongside other shared filters

Impact: en 655K→129K, pt-PT 1.5M→136K, fr 80K→56K (on reimport)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

internal/loader/prune.go

@@ -0,0 +1,36 @@
+package loader
+
+import (
+	"database/sql"
+	"fmt"
+	"log/slog"
+)
+
+// PruneOrphanWords removes words that have no evidence of being real:
+// no definitions, no synonyms, no antonyms, no translations,
+// no pronunciations, no etymology, no synset membership, and zero frequency.
+// These are typically mechanical affix expansion artifacts.
+type PruneOrphanWords struct{}
+
+func (PruneOrphanWords) Name() string { return "prune-orphans" }
+
+func (PruneOrphanWords) Load(db *sql.DB, _ string) error {
+	res, err := db.Exec(`
+		DELETE FROM words
+		WHERE frequency = 0
+		  AND NOT EXISTS (SELECT 1 FROM definitions   d  WHERE d.word_id  = words.id)
+		  AND NOT EXISTS (SELECT 1 FROM synonyms      s  WHERE s.word_id  = words.id)
+		  AND NOT EXISTS (SELECT 1 FROM antonyms      a  WHERE a.word_id  = words.id)
+		  AND NOT EXISTS (SELECT 1 FROM translations  t  WHERE t.word_id  = words.id)
+		  AND NOT EXISTS (SELECT 1 FROM pronunciations p WHERE p.word_id  = words.id)
+		  AND NOT EXISTS (SELECT 1 FROM etymology     e  WHERE e.word_id  = words.id)
+		  AND NOT EXISTS (SELECT 1 FROM word_synsets  ws WHERE ws.word_id = words.id)
+	`)
+	if err != nil {
+		return fmt.Errorf("prune orphans: %w", err)
+	}
+
+	n, _ := res.RowsAffected()
+	slog.Info("pruned orphan words", "deleted", n)
+	return nil
+}