The MWM can be performed in the afternoon, although I suggest that all testing be completed at least an hour before the colony room lights go off (or on, if you have them reversed). Similarly, I suggest that you don't start morning testing until at least an hour after they lights have turned on/off.
Our colony room lights are on from 6am to 6pm. To minimize circadian-related variability in behavior during testing, I typically test between 9am and 4pm, or if I'm using a cognitive assay that would be disrupted by noises from my coworkers, between 8pm and 3am.
Training in the morning and testing in the afternoon (or vice versa) should be fine. I would be cautious about training at night and testing in day (and vice versa), although it would still probably work.
The important parts are that the animals learn what they need to know, and that they are all in a similar circadian state when it's time for them to perform.