LC filters are a pain in this regard: you need to account for all the possible loads, in and out, and once calculated they stay that way. Therefore a change in either the input or the output needs recalculating the values.
Then you have an even order Chebyshev. As you probably know, these filters have the peculiarity of equiripples in the passband (or stopband for inverse kind). For active filters, this is not a problem, but passive filters are tricky: even orders don't have unity gain at DC (see this answer for more details), and that means there is a maximum value over which the response is no longer a Chebyshev.
But then, this line in your OP seems strange:
design a passive 4 pole filter for an active subwoofer.
If it's active, you don't need to use LC filters, use active filters which avoid the use of inductors. This is how active speakers are built: with a filter at the input, where the signal doesn't need power. Because, honestly, I don't know how you'll build those values in practice: 800 mH and 1.3 H are huge, unless you consider adding a magnetic core, but then there are all sorts of nonlinearities since this is audio, so you'll need an "air" core (i.e. no core). And having a 1 H inductance will take lots of turns, which means lots of parasitic capacitance, and space/weight.
So, rf-tools is tricky, therefore always verify the design with some SPICE tool (plenty free available, one right here on this site). And if you choose to go with passive LC filters, know that an LC filter is supposed to be lossless, therefore the attenuation is given by the resistive divider formed by the input and output resistances. Ideally, because in practice the inductors will have a resistance and capacitors, too, will contribute with their ESR.
