Automatically detect the content type retrieved by `fetch()`

[garysassano]

When interacting with a server that can respond with either JSON or an HTML document, is there a way to gracefully handle the response with minimal code? Specifically, I want to attempt parsing the response as JSON and, if that fails, fall back to treating it as plain text. Additionally, I would like the response.clone() to be handled automatically in the background.

Expected Behavior

• Attempt to parse the response as JSON.
• Fallback to plain text if JSON parsing fails.
• Automatically handle response.clone() to avoid issues with consuming the body multiple times.

[sholladay]

We're discussing something similar in #676. I suppose we could allow responseType to be an array to support fallbacks. There is a lot of overlap here with the Accept header. Would be nice to keep them in sync.