Use wasm-smith to canonicalize NaN in differential fuzzing (#3195)

* Update wasm-smith to 0.7.0

* Canonicalize NaN with wasm-smith for differential fuzzing

This then also enables floating point executing in wasmi in addition to
the spec interpreter. With NaN canonicalization at the wasm level this
means that we should be producing deterministic results between Wasmtime
and these alternative implementations.

fuzz/fuzz_targets/instantiate-swarm.rs

@@ -3,7 +3,7 @@
 use libfuzzer_sys::arbitrary::{Result, Unstructured};
 use libfuzzer_sys::fuzz_target;
 use std::time::Duration;
-use wasm_smith::{ConfiguredModule, SwarmConfig};
+use wasm_smith::{Module, SwarmConfig};
 use wasmtime::Strategy;
 use wasmtime_fuzzing::oracles::{self, Timeout};
 
@@ -30,7 +30,7 @@ fn run(data: &[u8]) -> Result<()> {
     config.memory64_enabled = u.arbitrary()?;
     // Don't generate modules that allocate more than 6GB
     config.max_memory_pages = 6 << 30;
-    let module = ConfiguredModule::new(config.clone(), &mut u)?;
+    let module = Module::new(config.clone(), &mut u)?;
 
     let mut cfg = wasmtime_fuzzing::fuzz_default_config(Strategy::Auto).unwrap();
     cfg.wasm_multi_memory(config.max_memories > 1);