Rethink getRecord() parsing records by content type #1067

getRecord() parses the body by the response content type: application/json is JSON.parsed, text types are decoded to a string, anything else stays a Buffer, with buffer: true as the only way out. It came up in a review discussion on #1051: since a string stored through setRecord() under contentType: 'application/json' now goes out verbatim, a non-JSON string saved that way fails to parse on the way back.

This issue is solutioning only. Decide whether the client should keep parsing by content type, and if so in what form:

• keep the current default and document the pitfall
• keep parsing but return the raw body when it fails to parse
• stop parsing by default and return the body with its content type

Compare with the Python client before choosing. Any change here is breaking.

✍️ Drafted by Claude Code