KoboldCpp client tweaks

[vegu-ai-tools]

• some issues with repetition breaking when open ai api is selected: rep_pen key error
• default visualizer to koboldcpp A1111 if koboldcpp is the client and has a sd model loaded
• default port should be 5001
• fix token count
• fix whatever the issue is with

  File "C:\AI\talemate\src\talemate\agents\base.py", line 85, in wrapper
    return await fn(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\AI\talemate\src\talemate\agents\narrator.py", line 46, in narration_wrapper
    response = await fn(self, *args, **kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\AI\talemate\src\talemate\agents\narrator.py", line 321, in narrate_query
    response = await Prompt.request(
               ^^^^^^^^^^^^^^^^^^^^^
  File "C:\AI\talemate\src\talemate\prompts\base.py", line 278, in request
    return await prompt.send(client, kind)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\AI\talemate\src\talemate\prompts\base.py", line 818, in send
    response = await client.send_prompt(str(self), kind=kind)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\AI\talemate\src\talemate\client\base.py", line 546, in send_prompt
    response, finalized_prompt = await self.auto_break_repetition(
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\AI\talemate\src\talemate\client\base.py", line 662, in auto_break_repetition
    prompt_param["max_tokens"] += pad_max_tokens
    ~~~~~~~~~~~~^^^^^^^^^^^^^^
KeyError: 'max_tokens'
 unhandled=True