cl-openid::parse-kv incorrectly splits message into key-value pairs by char code for #\Newline, instead of #\Newline character itself

[Avodonosov]

message elements (pairs) are separated from each other with #\Newline character.

cl-openid::parse-kv accepts the message as a byte array, and splits it into pairs by searching for byte 10 - char code for #\Newline.

This is incorrect, because 10 may also be a part of some multi-byte character.

We need to first decode byte array into string, and then split by #\Newlines.

---

The problem was detected because test cl-openid.parse-kv/random fails sometimes:

Sometimes this test case fails. For example see this log: ​https://cl-test-grid.appspot.com/blob?key=176038

But sometimes it passes; example: ​https://cl-test-grid.appspot.com/blob?key=171015

[avodonosov]

Fixed: ​https://github.com/cl-openid/cl-openid/commit/493152a333d3248db2996b7cbe3eb79cd844f357