Remove legacy wildcard key support

[robknight]

This PR removes legacy support for wildcard keys, which was interfering with the new integer "index" keys.

[ed255]

Thanks for finding this! I'm afraid that this fix is not enough.

This is the encoding of StatementArg:

pod2/src/middleware/statement.rs

Lines 568 to 571 in a2f427e

	/// Encoding:
	/// - None => `[0, 0, 0, 0, 0, 0, 0, 0]`
	/// - Literal(v) => `[[v], 0, 0, 0, 0]`
	/// - Key(root, key) => `[[root], [key]]`

Since the key in AnchoredKey can now be an integer, it could be 0. And then we can have an AnchoredKey that is encoded like a Literal.

The circuit uses this function to determine whether a StatementArg is a literal, and would treat an anchored key with key=0 as a literal

pod2/src/backends/plonky2/circuits/common.rs

Lines 1557 to 1562 in a2f427e

	fn statement_arg_is_value(&mut self, arg: &StatementArgTarget) -> BoolTarget {
	let zeros = iter::repeat(self.zero())
	.take(STATEMENT_ARG_F_LEN - VALUE_SIZE)
	.collect::<Vec<_>>();
	self.is_equal_slice(&arg.elements[VALUE_SIZE..], &zeros)
	}

We need to figure out a way to distinguish the encoding of StatementArg::Literal and StatementArg::Key with ak.key = integer.
One option is to add a new field element to store a tag. But will affect performance, as it grows the number of fields required to encode a statement :(

Perhaps changing the encoding of integers isn't a bad idea after all.

Update: I've just seen you opened an issue about this #513