Improve simple training example in v3 migration (#6438)

* Create the examples once
* Use the examples in the initialization
* Provide the batch size
* Fix `begin_training` migration example

website/docs/usage/v3.md

@@ -969,18 +969,18 @@ The [`Language.update`](/api/language#update),
 raw text and a dictionary of annotations.
 
 ```python
-### Training loop {highlight="11"}
+### Training loop {highlight="5-8,12"}
 TRAIN_DATA = [
     ("Who is Shaka Khan?", {"entities": [(7, 17, "PERSON")]}),
     ("I like London.", {"entities": [(7, 13, "LOC")]}),
 ]
-nlp.initialize()
+examples = []
+for text, annots in TRAIN_DATA:
+    examples.append(Example.from_dict(nlp.make_doc(text), annots))
+nlp.initialize(lambda: examples)
 for i in range(20):
-    random.shuffle(TRAIN_DATA)
+    random.shuffle(examples)
-    for batch in minibatch(TRAIN_DATA):
+    for batch in minibatch(examples, size=8):
-        examples = []
-        for text, annots in batch:
-            examples.append(Example.from_dict(nlp.make_doc(text), annots))
         nlp.update(examples)
 ```
 
@@ -995,7 +995,7 @@ network,
 setting up the label scheme.
 
 ```diff
-- nlp.initialize(examples)
+- nlp.begin_training()
 + nlp.initialize(lambda: examples)
 ```