Small oversight on the reverse bilinear transform when no lowcut filter is applied

[louisreine]

Since applying this bilinear transform before

b = 2 / Ts * npa([1.0, -1.0])  # don't need Ts scaling but simpler to keep for receiver post-processing
a = npa([1.0, 1.0])

The correct reverse operation should be :

b = Ts / 2 * npa([1, 1])
a = npa([1, -1])

Btw, I didn't get why the bilinear transform helps with single precision operation. Is there a signal processing reason behind this method ?

Thanks for all of your work !