fix: use proper max bit size during truncation

compiler/noirc_evaluator/src/acir/mod.rs

@@ -1965,7 +1965,7 @@ impl<'a> Context<'a> {
         &mut self,
         value_id: ValueId,
         bit_size: u32,
-        max_bit_size: u32,
+        mut max_bit_size: u32,
         dfg: &DataFlowGraph,
     ) -> Result<AcirVar, RuntimeError> {
         assert_ne!(bit_size, max_bit_size, "Attempted to generate a noop truncation");
@@ -1986,6 +1986,7 @@ impl<'a> Context<'a> {
                     let integer_modulus = power_of_two::<FieldElement>(bit_size);
                     let integer_modulus = self.acir_context.add_constant(integer_modulus);
                     var = self.acir_context.add_var(var, integer_modulus)?;
+                    max_bit_size += 1;
                 }
             }
             Value::Param { .. } => {

test_programs/execution_failure/regression_7759/stdout.txt

@@ -0,0 +1 @@
+[regression_6834] Circuit output: Field(0)

test_programs/execution_success/regression_7962/Nargo.toml

@@ -0,0 +1,6 @@
+[package]
+name = "regression_7962"
+type = "bin"
+authors = [""]
+
+[dependencies]

test_programs/execution_success/regression_7962/Prover.toml

@@ -0,0 +1 @@
+"v0" = "18446744073709551615"

test_programs/execution_success/regression_7962/src/main.nr

@@ -0,0 +1,4 @@
+fn main(v0: u64) -> pub u64 {
+    let v1 = v0 - 1;
+    v1 & 1
+}

test_programs/execution_success/regression_7962/stdout.txt

@@ -0,0 +1 @@
+[regression_7962] Circuit output: Field(0)