core/vdbe: Micro-optimize "zero_or_null" opcode

It's a hot instruction for TPC-H, for example, so worth optimizing.
Reduces op_zero_or_null() from 5.6% to 2.4% in CPU flamegraph for TCP-H
Q1.

core/vdbe/execute.rs

@@ -6667,9 +6667,7 @@ pub fn op_zero_or_null(
     mv_store: Option<&Arc<MvStore>>,
 ) -> Result<InsnFunctionStepResult> {
     load_insn!(ZeroOrNull { rg1, rg2, dest }, insn);
-    if *state.registers[*rg1].get_value() == Value::Null
-        || *state.registers[*rg2].get_value() == Value::Null
-    {
+    if state.registers[*rg1].is_null() || state.registers[*rg2].is_null() {
         state.registers[*dest] = Register::Value(Value::Null)
     } else {
         state.registers[*dest] = Register::Value(Value::Integer(0));

core/vdbe/mod.rs

@@ -229,6 +229,13 @@ pub enum Register {
     Record(ImmutableRecord),
 }
 
+impl Register {
+    #[inline]
+    pub fn is_null(&self) -> bool {
+        matches!(self, Register::Value(Value::Null))
+    }
+}
+
 /// A row is a the list of registers that hold the values for a filtered row. This row is a pointer, therefore
 /// after stepping again, row will be invalidated to be sure it doesn't point to somewhere unexpected.
 #[derive(Debug)]