fix: encode_text latency

Switching the condition `if not self.tokenizer`which takes 20~30ms to an explicit `if self.tokenizer is None:` ( < 1ms)

modeling_clip.py

@@ -329,7 +329,7 @@ class JinaCLIPModel(JinaCLIPPreTrainedModel):
         self.post_init()
 
     def get_tokenizer(self):
-        if not self.tokenizer:
+        if self.tokenizer is None:
             self.tokenizer = AutoTokenizer.from_pretrained(
                 self.config._name_or_path, trust_remote_code=True
             )