I.e. instead of `BytesIO(b"abc").read()` returning `b"abc"`, it returns only `b"a"`, requiring multiple calls to get the full buffer. This is allowed, cf. https://docs.python.org/3.14/library/io.html#io.RawIOBase.read. Follow-up on https://github.com/smheidrich/py-json-stream-rs-tokenizer/pull/105 Related to https://github.com/smheidrich/py-json-stream-rs-tokenizer/issues/107