Optimize codegen slightly calling wasm functions

Currently wasm-calls work with `Result<T, Trap>` internally but `Trap`
is an enum defined in `wasmtime-runtime` which is actually quite large.
Since traps are supposed to be rare this commit changes these functions
to return a `Box<Trap>` which is un-boxed later up in the `wasmtime`
crate within a `#[cold]` function.

crates/wasmtime/src/engine.rs

@@ -87,7 +87,7 @@ impl Engine {
     /// latency of WebAssembly calls are extra-important, which is not
     /// necessarily true of all embeddings.
     pub fn tls_eager_initialize() -> Result<(), Trap> {
-        wasmtime_runtime::tls_eager_initialize().map_err(Trap::from_runtime)
+        wasmtime_runtime::tls_eager_initialize().map_err(Trap::from_runtime_box)
     }
 
     /// Returns the configuration settings that this engine is using.

crates/wasmtime/src/func.rs

@@ -1056,7 +1056,7 @@ pub(crate) fn invoke_wasm_and_catch_traps<T>(
         );
         exit_wasm(store, exit);
         store.0.entering_native_hook()?;
-        result.map_err(Trap::from_runtime)
+        result.map_err(Trap::from_runtime_box)
     }
 }
 

crates/wasmtime/src/trap.rs

@@ -161,6 +161,11 @@ impl Trap {
         )
     }
 
+    #[cold] // see Trap::new
+    pub(crate) fn from_runtime_box(runtime_trap: Box<wasmtime_runtime::Trap>) -> Self {
+        Self::from_runtime(*runtime_trap)
+    }
+
     #[cold] // see Trap::new
     pub(crate) fn from_runtime(runtime_trap: wasmtime_runtime::Trap) -> Self {
         match runtime_trap {