Fix PyFloat::py_new always returning new float object issue (#3979)

extra_tests/snippets/builtin_float.py

@@ -441,6 +441,8 @@ assert float('inf').hex() == 'inf'
 assert float('-inf').hex() == '-inf'
 assert float('nan').hex() == 'nan'
 
+assert float(math.nan) is float(math.nan)
+
 # Test float exponent:
 assert 1 if 1else 0 == 1
 

vm/src/builtins/float.rs

@@ -142,6 +142,10 @@ impl Constructor for PyFloat {
         let float_val = match arg {
             OptionalArg::Missing => 0.0,
             OptionalArg::Present(val) => {
+                if cls.is(vm.ctx.types.float_type) && val.class().is(vm.ctx.types.float_type) {
+                    return Ok(val);
+                }
+
                 if let Some(f) = val.try_float_opt(vm)? {
                     f.value
                 } else {