there is some code duplication but not sure if there is a cleaner way of deduplicating this

Could we extract it to a common function that returns length for decrypted content buffer? This particular example seems to be too much duplication.

I think it wouldn't hurt to replace it with constructor that doesn't require casting:

std::string{decryptedMessage.begin(), decryptedMessage.end()};

not sure about it - this function will only call olm_decrypt_max_plaintext_length which is weird because it's like creating a wrapper over direct olm function call

D11836