IIRC, the LPF is in the open-loop path. When NFB is applied, it results in rise-time increase. The low-level HF response is almost unchanged because of NFB compensating, but as soon as the frequency/level combination results in slew-rate limiting, it can be legitimately be qualified as some sort of rise-time control.
I discovered that a long time ago, when my opamp of choice was the 748. Accidentally, the compensation cap (typically 30 pF) was replaced with a higher value in production (300 or 3000pF). Standard QC did not see that but final audible test was suspicious. It took some time to find out, because the effect is quite subtle in normal operating conditions, where energy tends to decrease at HF.
Had we done the test on cymbals or soprano voice, it would probably have been more obvious...
Finally, a full frequency sweep at high-level revealed the issue, with the output mutating from a standard sinewave to a near-perfect triangle.