[onert/train] Fix value error of MaxPool2d with stride

[zetwhite]

What

MaxPool2D (stride > 1) gets different values compared to tensorflow.

• tensorflow https://gist.github.com/zetwhite/34e9fb14968d9ee1bc4e4696f157bd9c

• onert test cases https://github.com/ragmani/ONE/blob/3920c734227f8f5aa18cfd994293abb41363f447/tests/nnfw_api/src/GenModelTests/nontrainable_op_trains/MaxPool2D.test.cc#L148-L203

[zetwhite]

Ah, There is no value dfference between onert train and tensorflow.. I misunderstood the situation.

This is is resolved - i checked with this commit (https://github.com/Samsung/ONE/pull/13253/commits/6f5c6f03bbf71c222633cf1ce30bd3d9b99350cf)