Debugging: avoid re-patching code on single-step update when state doesn't change. (#12638)

When the setter for the single-step flag is given an `enable` boolean
that matches the existing state (i.e., enabling when already enabled or
disabling when already disabled), we should not loop through and
re-patch all breakpoint patch sites. This is an optimization, not
required for correctness, but a rather obvious optimization to make (and
important when building higher-level APIs for e.g. a single-step command
that always set the flag).

Author: cfallin

crates/wasmtime/src/runtime/debug.rs

@@ -1088,6 +1088,11 @@ impl<'a> BreakpointEdit<'a> {
             "single_step({enabled}) with breakpoint set {:?}",
             self.state.breakpoints
         );
+        if self.state.single_step == enabled {
+            // No change to current state; don't go through the effort of re-patching and
+            // re-publishing code.
+            return Ok(());
+        }
         let modules = self.registry.all_modules().cloned().collect::<Vec<_>>();
         for module in modules {
             let mem = Self::get_code_memory(self.registry, &mut self.dirty_modules, &module)?;