Do we need to track max_bit_size in Instruction::Truncate?

[TomAFrench]

Instruction::Truncate compiles down to a euclidean division where max_bit_size is passed in as the bit_size argument. The only place this is used in the circuit are

noir/compiler/noirc_evaluator/src/acir/acir_variable.rs

Lines 826 to 838 in 8d7f1f9

	// maximum bit size for q and for [r and rhs]
	let mut max_q_bits = bit_size;
	let mut max_rhs_bits = bit_size;
	// when rhs is constant, we can better estimate the maximum bit sizes
	if let Some(rhs_const) = rhs_expr.to_const() {
	max_rhs_bits = rhs_const.num_bits();
	if max_rhs_bits != 0 {
	if max_rhs_bits > bit_size {
	return Ok((zero, zero));
	}
	max_q_bits = bit_size - max_rhs_bits + 1;
	}
	}

where we explicitly ignore this if the rhs term of the division is known at compile time (as is the case for truncations), also in

noir/compiler/noirc_evaluator/src/acir/acir_variable.rs

Lines 845 to 846 in 8d7f1f9

	AcirValue::Var(lhs, AcirType::unsigned(bit_size)),
	AcirValue::Var(rhs, AcirType::unsigned(bit_size)),

Where we're just specifying the types to be given to the brillig VM (where there's no difference between passing the real bit sizes and fake ones as long as they're big enough)

Brillig ignores max_bit_size entirely.

This raises the question of whether we just remove this field from Instruction::Truncate and pass in 254 as the bit size during ACIR gen.