First of all, the comment you link to talks about the format of the ECDSA signature, not the keypair. Secondly, it is a bit misleading:
The output from the ECDSA algorithm is two integers in the interval [1, n-1]. The ECDSA standard (FIPS 183-3) does not specify way a standard method to encode this pair of numbers as a array of bytes.
One way is by encapsulating the numbers in an ASN.1 SEQUENCE. This is the way specified by ANSI X9.62 and RFC3278. It is the standard output from Java and (AFAIR) Microsoft CNG/.NET.
Another way is by left-padding the numbers with zeroes so they have the same byte-length as n
and then just concatenating them. This is done by PKCS#11 and most smartcard implementations.
If I read the source code correctly, the SJCL encodes the ECDSA signature in the second way. You can easily convert this format to the first one.