What you seem to be missing is that the post is specifically about conversion between floating points encoded in decimal and floating point encoded in binary. BCD are decimal digits encoded as binary, they're not binary-based floating point.
It is made clear from the very first sentence of the article that it is about representations that will round-trip between number bases, not whether or not it is possible to encode decimal floating point numbers with two digit accuracy in 7 bits.
It is made clear from the very first sentence of the article that it is about representations that will round-trip between number bases, not whether or not it is possible to encode decimal floating point numbers with two digit accuracy in 7 bits.