I was told this is the reason why Airbus doesn't recommend using APU bleed during takeoff (rather packs off instead), because it's not ensured crew wouldn't forget to turn it off in climb.
The story from my Toulouse days was to think what happens should the APU fail and the engines' bleed start to supply the packs - effects on both performance and N1 / EGT.
Both have a merit.
That would follow a certain philosophy of SOP design, that they need to cater for the next failure. This is also a nice test to try when somebody starts to feel they are smarter than OEM FCOM.
