If it was "official" guidance it would be worded a LOT better than that. Openreach documents need a degree or two to translate, which I suppose could be why the engineer got it completely messed up.
If there is any truth in that at all the figures are reversed. You cap (band) the sync rate and then set the profile rate below that to account for overheads. It wouldn't be unusual to see your sync rate 20Mbit below the attainable rate when on interleaving, but that's not a cap, its miscalculation of the attainable rate (as the attainable rate is entirely an educated guess by the DSLAM/modem based on your line statistics which themselves are likely averaged).
Also, on full-sync you will be capped about 4Mbit below the sync rate (its actually a percentage of your sync rate thats different depending on fastpath or interleaved), there is no such thing as an uncapped line that I'm aware of as that would allow bursts of traffic dramatically exceeding your bandwidth to reach the cabinet and then have to be dropped, which would be a huge waste of backhaul capacity and cause horrific latency spikes. The entire purpose of the various error correction schemes is to try to minimise data needing to be sent from the ISP more than once, better it be re-transmitted by the line card from its own buffer at a much lower level in the network stack - the ISP need never know.
While its not impossible, I'm just not convinced Openreach would go to the trouble of having an additional stability profile for ECI as that's the whole point of having DLM to tweak the settings within the performance or stability profile set. Its a lot more effort for them and the ISP to have different options for different cabinets, or even on the same cabinet (my cabinet is ECI with a Huawei expansion pod).
Also of note, my recent ECI DLM reset put me straight on fastpath and it seem synced solid for nearly three months now.