They want, e.g. "lots of wodget-sellers jailed so that honorable wodget-avoiders don't have to deal with the wodget menace". With the insurance mechanism, however, they can't get that by itself; at best, they can get that plus a large transfer of wealth from the "good guys" to the "bad guys" -- which is much less politically appealing.

If the ban on wodgets is expected to be binding, then a proposed ban on wodgets with probability p of being enacted does decrease the expected value of wodget-selling capital. Wodget-sellers can choose whether to cut their losses by taking out insurance, keep their risk exposure unchanged or even reverse their exposure by betting more than their existing interest--but they cannot get a free lunch unless they have magical foreknowledge of future events.

Yes, to the extent that wodget-sellers trade their risk with wodget-avoiders, the political positions of either will shift towards neutrality. But this will hardly make legislatures useless.

As for your suggestion that having more money could help against enforcement, well, let's put it in decision-theoretical terms: do you honestly think that transferring money from the possible world in which selling wodgets is legal to the possible world in which doing so is illegal is actually a sound investment? Because this is what the transaction boils down to.

A genuine problem with political prediction markets is that niche speculators can rent-seek by manipulating the legislation process. I don't have a real answer to this, but it seems to be a self-limiting problem: e.g. a speculator who tries to manipulate the legislature into enacting some law and fails has wasted effort in addition to losing his bet.