OpenAIClientV1 does not implement retries

[mlevtov]

Describe the bug
Retries for transient errors are implemented for OpenAIClientV0, but not for OpenAIClientV1

To Reproduce
Steps to reproduce the behavior:

import guardrails as gd
import openai
from multiprocessing.pool import ThreadPool as Pool

text_chunk = <Chunk of text that’s ~4k tokens>

def call_gd(text_chunk):
  guard = gd.Guard(prompt="summarize this chunk of text: ${chunk}", num_reasks=5)
  result = guard(
      openai.chat.completions.create,
      prompt_params={"chunk": text_chunk},
      max_tokens=1024,
      temperature=0.1
  )

with Pool(16) as p:
  results = list(p.imap_unordered(call_gd, text_chunk))

Expected behavior
Calls should retry when receiving a RateLimitError.

Library version:
guardrails-ai==0.4.1
openai==1.13.3

Additional context
See this conversation