Support channel first data format

[angerhang]

Currently, we only support data format: N x M x 3. We would also ideally also support N x 3 x M by adding a flag in the extract_features function:

feats = extract_features(data, feature_set='full', sample_rate=sample_rate)

[angerhang]

def chanfirst2chanlast(my_data):
    x = my_data[:, 0, :]
    y = my_data[:, 1, :]
    z = my_data[:, 2, :]


    x = x.reshape(len(x), -1, 1)
    y = y.reshape(len(y), -1, 1)
    z = z.reshape(len(z), -1, 1)

    final_data = np.concatenate((x, y, z), axis=2)
    return final_data

Something like this should do.