SK-3061: fix Bulk Insert README's Token snippet to actually iterate the response (#411)

The snippet added when getTokens() was made typed (#408) used `record`
without ever introducing it - it reads as if it stood alone, but record
only exists inside a loop over insertResponse.getRecords(). Wrapped it in
that outer loop (plus a null check, since getTokens() is null on a failed
record), matching the pattern the Bulk Detokenize section's equivalent
metadata snippet already follows.

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

Author: Devesh-Skyflow

flowvault/README.md

@@ -485,8 +485,12 @@ Sample response:
 `getTokens()` returns `Map<String, List<Token>>` — one entry per token group configured on that column, so a column with a single token group still comes back as a one-element list, not a bare string. On the wire the API models this generically (`Object`, not a fixed type) to stay flexible, but the SDK parses it into `Token` objects before handing it back, so callers get `Token.getToken()`/`Token.getTokenGroupName()` directly with no casting required:
 
 ```java
-for (Token token : record.getTokens().get("card_number")) {
-    System.out.println(token.getTokenGroupName() + " -> " + token.getToken());
+for (BulkInsertResponseRecord record : insertResponse.getRecords()) {
+    if (record.getTokens() != null) {
+        for (Token token : record.getTokens().get("card_number")) {
+            System.out.println(token.getTokenGroupName() + " -> " + token.getToken());
+        }
+    }
 }
 ```