fix: correct context pruning order and vector alignment

- Prune context BEFORE processing (LSTM forget gate pattern)
  so LLM only sees relevant context, instead of pruning after the fact
- Fix ensureTrained() to recompute all event vectors after retraining
  vectorizer, fixing feature-space mismatch between stored vectors and
  query vector that made relevance scoring effectively random
- Add read lock to knowledge BuildTree() (data race fix)
- Log writeIndex() errors instead of discarding them
- Fix TOCTOU race in document ContextToDoc() dedup
- Fix healthcheck timing (measure elapsed before cleanup)
- Refactor waiter CLI into separate files (state, conn, config, editor,
  history, builtin) for maintainability
- Add CLI plugin API key authentication

internal/plugins/healthcheck/plugin.go

@@ -456,8 +456,12 @@ func (p *Plugin) testKnowledgeRaw() checkResult {
 	}
 
 	results := hcKnowledge.Search("健康检查测试标记", 3)
+
 	elapsed := time.Since(start)
 
+	// 清理测试条目，避免积累
+	hcKnowledge.Remove(marker)
+
 	if len(results) > 0 {
 		return checkResult{
 			Name:   "knowledge",