EdgeNeXt: Efficiently Amalgamated CNN-Transformer Architecture for Mobile Vision Applications