Bryston uses two diff-amps, (long-tailed pairs), at the input.
We deliberately do not use constant-current sources for the emitters because it is not necessary to reduce second harmonic as it is automatically cancelled anyway.
However, constant-current sources introduce higher-order harmonics due to the junction capacitance in the active components.
I am afraid the result was from the limitation of the particular CCS used in the Bryton lab.
There are variety of CCS designs, hence different characteristics.
Different implementation produces different result.
The listeners testimonials are pointing different direction than the Bryston's.