input_dscaled_by is passed in nn.py but unused in FILM layer

[xvdp]

Line 85 of nn.py passes

FiLM(...
     input_dscaled_by=np.product(film_factors[:i+1])  # for proper positional encodings initialization`
)

is passed into linear_modulation.py

class FeatureWiseLinearModulation(BaseModule):
    def __init__(self, in_channels, out_channels, input_dscaled_by):

but never used
does it mean that the positional encoding is misaligned wrt. channels or is this simply code that needs linting? I looked through various implementations of FiLM generation layers and I did not find similar references to the down scaling

Thank you
xvdp