Models for sequence data need 1D variants of the convolutional and pooling layers. AC: [WhistleNetMk16](https://github.com/NaoDevils/AudioProcessing/blob/61ea50dfd5acb66e53d50fea16ae2765633ec904/PreTrainedNetworks/WhistleDetection/WhistleNetMk16.h5) can be run in CompiledNN