Implications of Using AccountId::default() as parent AccountId for Location-Conversion

[mustermeiszer]

Given a note on the default behaviour of AccountId32

Quote:

We recently discovered that anyone can sign transactions in Substrate with a zero account (AccountId::default()). The reason behind this is that cryptographically a private key of zero makes to an actual complete curve in sr25519 and ed255519. This is coming directly from schnorrkel.

Resolved for substrate via paritytech/substrate#10403

Aren't the following lines a bit dangerous, given that many parachains will most likely plug in the normal AccountId32 type there?

polkadot/xcm/xcm-builder/src/location_conversion.rs

Lines 40 to 59 in 5a7004f

	pub struct ParentIsDefault<AccountId>(PhantomData<AccountId>);
	impl<AccountId: Default + Eq + Clone> Convert<MultiLocation, AccountId>
	for ParentIsDefault<AccountId>
	{
	fn convert_ref(location: impl Borrow<MultiLocation>) -> Result<AccountId, ()> {
	if location.borrow().contains_parents_only(1) {
	Ok(AccountId::default())
	} else {
	Err(())
	}
	}
	fn reverse_ref(who: impl Borrow<AccountId>) -> Result<MultiLocation, ()> {
	if who.borrow() == &AccountId::default() {
	Ok(Parent.into())
	} else {
	Err(())
	}
	}
	}

[bkchr]

CC @KiChjang

[KiChjang]

Yeah, we should remove that, I guess this just wasn't done in the polkadot repo yet.