chat_github crashes on rate limit (429) due to invalid JSON body

[D-M4rk]

When using chat_github or other OpenAI-compatible providers and/or spamming with chat_parallel, hitting a rate limit (429 Too Many Requests) causes ellmer to crash with a JSON parsing error instead of retrying or reporting the error.

This occurs because the GitHub Models endpoint returns a Content-Type: application/json header but a plain text body "Too many requests... "etc. when rate limited... i.e. The base_request_error method attempts to parse this as JSON and fails.

Reproduction code

library(ellmer)
# Use a small model to hit rate limits quickly
chat <- chat_github(model = "gpt-4o-mini")
prompts <- as.list(paste("Say hello", 1:50))

# Trigger 429
tryCatch(
  parallel_chat(chat, prompts, max_active = 20),
  error = function(e) print(e)
) 

Proposed Fix-
Update base_request_error in R/provider-openai-compatible.R to wrap resp_body_json() in a tryCatch. If JSON parsing fails, fall back to returning the raw string body.
This allows httr2 to correctly identify the 429 error and apply retry logic.

I'll attempt a concise pull request...