I just tried IDAC on CH1 and didn't have any problems. After calibrating, I power-cycled and went back and verified the results at 0.1A intervals. Everything was reasonably close from 0.1A to 3.2A.
I think they have fixed the bug introduced in 1.11 but there is another bug that was there before. If you start a calibration and exit out, you have to power-cycle the device before starting again. If you just go back into calibration, the results will be wrong. As a test, I started IDAC cal and measured 0.070A for the first step and entered that value. Then I exited cal and went back in (password 2012, etc.) and reselected IDAC. This time the first step measured 0.100A. I think what is happening is that the first time you enter calibration for a channel, they clear out the previous calibration settings so you get the raw hardware response. If you try to repeat a device calibration with out power-cycling, they don't clear the old cal data and the power supply outputs values that are adjusted by the old cal data. You could probably fix that by issuing the ":CALibration:Clear" SCPI command. (that's the command that's no longer mentioned in the latest Rigol calibration document.
Normally, you wouldn't want to exit out of a calibration once you start, but I find it's easy to do accidentally. When selecting which device to calibrate you have to press the VDAC, IDAC, VADC, or IADC button and then press the back arrow to get back to the main cal menu. But when entering values and you want to change the +/- sign, if you press the back arrow after selecting +/-, it kicks you completely out of calibration mode. That has bitten me several times. If you don't power-cycle before starting again, the results will be wrong.
When in calibration mode, they should never correct the output using the cal data tables. They try to achieve this by clearing out the tables when you enter calibration for the first time. That is not good enough. If you restart calibration or go back and repeat a cal step you've already done, the previous cal data is not blank and it will throw off the results.
I fear that it's absolutely hopeless trying to explain that to Rigol.