My experience is that the current limiting accuracy (on the main, software set, current limit) isn't great. +/- 3 to 5 amps is about as good as they get. The main issue seems to be variation in resistance of the whole shunt current path, including how much solder there is on the board 0V trace. I don't really think the current limit needs to be much better than this though, as it's only intended to be a rough setting - I doubt most people would be able to tell the difference between 20A and 25A unless they were looking for it.
Setting the phase current for 2.5 times the current limit is OK, but if you are trying to pull 25A from this controller in standard form then it will get warm unless you change the FETs. As standard, they come with either 60NF06 FETs (16 mohm Rdson) or sometimes 75NF75 FETs (13 mohm Rdson). These are quite lossy devices, for example, the IRFB3077 FETs that I'm running in one of these have an Rdson of typically 2.8 mohm. This makes a big difference to how hot they get. For example, 25A through a 60NF06 FET will create around 10 watts of heat (per FET) whereas 25A through an IRFB3077 FET will create about 1.75 watts of heat. The IRFB3077 FETs are good for around 65 to 70V working voltage (they are rated at 75V max) and are a good choice for this controller. Their max current rating is 210A for the silicon, but restricted to 120A by the package. In practice I'd have no qualms at running these FETs at 50A or so in a 6 FET controller, maybe more if the use was such that the controller wasn't being held at high current for long periods of time. A 60V pack with a 50A current limit on one of these controllers fitted with these FETs would be OK for around 3kW peak power, pretty impressive for such a small unit. Change the capacitors to 100V ones (they are usually 63V) and change the FETs for IRFB4110s (100V, about 3.7 mohm Rdson) and you could probably see up to 5kW or so.