Use a SmallVec for ABIArgSlots (#4586)

These are always length 1 for Wasm benchmarks.

<h3>Sightglass Benchmark Results</h3>

```
compilation :: nanoseconds :: benchmarks/spidermonkey/benchmark.wasm

  Δ = 328624015.86 ± 40274677.93 (confidence = 99%)

  main.so is 0.88x to 0.91x faster than slots-smallvec.so!
  slots-smallvec.so is 1.10x to 1.13x faster than main.so!

  [3070752447 3203778792.55 3446269274] main.so
  [2503544039 2875154776.69 3197966713] slots-smallvec.so

compilation :: nanoseconds :: benchmarks/pulldown-cmark/benchmark.wasm

  Δ = 9685705.06 ± 3221286.87 (confidence = 99%)

  main.so is 0.91x to 0.96x faster than slots-smallvec.so!
  slots-smallvec.so is 1.05x to 1.09x faster than main.so!

  [129356493 145594942.79 165038803] main.so
  [118555011 135909237.73 188780619] slots-smallvec.so

compilation :: nanoseconds :: benchmarks/bz2/benchmark.wasm

  No difference in performance.

  [79080493 86757564.46 112649639] main.so
  [78083384 85934125.69 94992743] slots-smallvec.so
```

cranelift/codegen/src/isa/aarch64/abi.rs

@@ -212,7 +212,7 @@ impl ABIMachineSpec for AArch64MachineDeps {
                     let upper_reg = xreg(next_xreg + 1);
 
                     ret.push(ABIArg::Slots {
-                        slots: vec![
+                        slots: smallvec![
                             ABIArgSlot::Reg {
                                 reg: lower_reg.to_real_reg().unwrap(),
                                 ty: param.value_type,

cranelift/codegen/src/isa/x64/abi.rs

@@ -125,7 +125,7 @@ impl ABIMachineSpec for X64ABIMachineSpec {
                 );
             }
 
-            let mut slots = vec![];
+            let mut slots = ABIArgSlotVec::new();
             for (rc, reg_ty) in rcs.iter().zip(reg_tys.iter()) {
                 let intreg = *rc == RegClass::Int;
                 let nextreg = if intreg {

cranelift/codegen/src/machinst/abi_impl.rs

@@ -153,6 +153,11 @@ impl ABIArgSlot {
     }
 }
 
+/// A vector of `ABIArgSlot`s. Inline capacity for one element because basically
+/// 100% of values use one slot. Only `i128`s need multiple slots, and they are
+/// super rare (and never happen with Wasm).
+pub type ABIArgSlotVec = SmallVec<[ABIArgSlot; 1]>;
+
 /// An ABIArg is composed of one or more parts. This allows for a CLIF-level
 /// Value to be passed with its parts in more than one location at the ABI
 /// level. For example, a 128-bit integer may be passed in two 64-bit registers,
@@ -169,7 +174,7 @@ pub enum ABIArg {
     /// parts used to store a value of this type.
     Slots {
         /// Slots, one per register part.
-        slots: Vec<ABIArgSlot>,
+        slots: ABIArgSlotVec,
         /// Purpose of this arg.
         purpose: ir::ArgumentPurpose,
     },
@@ -206,7 +211,7 @@ impl ABIArg {
         purpose: ir::ArgumentPurpose,
     ) -> ABIArg {
         ABIArg::Slots {
-            slots: vec![ABIArgSlot::Reg { reg, ty, extension }],
+            slots: smallvec![ABIArgSlot::Reg { reg, ty, extension }],
             purpose,
         }
     }
@@ -219,7 +224,7 @@ impl ABIArg {
         purpose: ir::ArgumentPurpose,
     ) -> ABIArg {
         ABIArg::Slots {
-            slots: vec![ABIArgSlot::Stack {
+            slots: smallvec![ABIArgSlot::Stack {
                 offset,
                 ty,
                 extension,