YJIT: A64: Use ADDS/SUBS/CMP (immediate) when possible

We were loading 1 into a register and then doing ADDS/SUBS previously.
That was particularly bad since those come up in fixnum operations.

```diff
   # integer left shift with rhs=1
-  mov x11, #1
-  subs x11, x1, x11
+  subs x11, x1, #1
   lsl x12, x11, #1
   asr x13, x12, #1
   cmp x13, x11
-  b.ne #0x106ab60f8
-  mov x11, #1
-  adds x12, x12, x11
+  b.ne #0x10903a0f8
+  adds x12, x12, #1
   mov x1, x12
```

Note that it's fine to cast between i64 and u64 since the bit pattern is
preserved, and the add/sub themselves don't care about the signedness of
the operands.

CMP is just another mnemonic for SUBS.

yjit/src/asm/arm64/mod.rs

@@ -276,6 +276,9 @@ pub fn cmp(cb: &mut CodeBlock, rn: A64Opnd, rm: A64Opnd) {
 
             DataReg::cmp(rn.reg_no, rm.reg_no, rn.num_bits).into()
         },
+        (A64Opnd::Reg(rn), A64Opnd::Imm(imm12)) => {
+            DataImm::cmp(rn.reg_no, (imm12 as u64).try_into().unwrap(), rn.num_bits).into()
+        },
         (A64Opnd::Reg(rn), A64Opnd::UImm(imm12)) => {
             DataImm::cmp(rn.reg_no, imm12.try_into().unwrap(), rn.num_bits).into()
         },

yjit/src/backend/arm64/mod.rs

@@ -315,7 +315,11 @@ impl Assembler
             match opnd {
                 Opnd::Reg(_) | Opnd::CArg(_) | Opnd::InsnOut { .. } => opnd,
                 Opnd::Mem(_) => split_load_operand(asm, opnd),
-                Opnd::Imm(_) => asm.load(opnd),
+                Opnd::Imm(imm) => if ShiftedImmediate::try_from(imm as u64).is_ok() {
+                    opnd
+                } else {
+                    asm.load(opnd)
+                }
                 Opnd::UImm(uimm) => {
                     if ShiftedImmediate::try_from(uimm).is_ok() {
                         opnd
@@ -1704,4 +1708,20 @@ mod tests {
             0x8: csel x1, x11, x12, lt
         "});
     }
+
+    #[test]
+    fn test_add_with_immediate() {
+        let (mut asm, mut cb) = setup_asm();
+
+        let out = asm.add(Opnd::Reg(TEMP_REGS[1]), 1.into());
+        let out = asm.add(out, 1_usize.into());
+        asm.mov(Opnd::Reg(TEMP_REGS[0]), out);
+        asm.compile_with_num_regs(&mut cb, 2);
+
+        assert_disasm!(cb, "2b0500b16b0500b1e1030baa", {"
+            0x0: adds x11, x9, #1
+            0x4: adds x11, x11, #1
+            0x8: mov x1, x11
+        "});
+    }
 }