There seems to be considerable debate about whether it is better to increase drag temporarily by raising the gear, or not - maybe this issue is deserving of simulation as an answer cannot be arrived at intuitively?
It's a simple calculation in a performance deck if you have access to somebody who knows
...and can run the figures with 2/3 of the LEDs retracted...
There is the possibility that they might have needed to accelerate in ground effect before climbing, given that they were close enough to max AoA in the config. they found themselves in to trigger a stick shake. A few seconds thought (which is all they had) might have led to the conclusion that this would be easier/more effective with the wheels up?
In the windshear and gpws situation greater minds have determined that there should be no change in configuration until the warnings have stopped and you're safely climbing away, gear doors produce drag...so 99% of the above is...
I think leaving the gear down in those scenarios is more to do with possible ground contact during the recovery than with drag. If you give the terrain a glancing blow with the undercarriage rather than the engine pods, it may produce a better outcome. Keeping the wing the same section allows for a better initial ROC in the GPWS instance and a better climb rate/gradient in windshear.