Fix cast_sign_loss false positive for float to uint casts

The cast_sign_loss lint was incorrectly triggering on float to unsigned
integer casts, which cannot actually lose sign information since floats
don't have the same sign semantics as signed integers.

This change modifies the should_lint function to exclude float to uint
casts from the cast_sign_loss lint.

changelog: Fix [] false positive for float to unsigned integer casts

fixed tests

Author: krikera

clippy_lints/src/casts/cast_sign_loss.rs

@@ -76,9 +76,8 @@ fn should_lint<'cx>(cx: &LateContext<'cx>, cast_op: &Expr<'_>, cast_from: Ty<'cx
             true
         },
 
-        (false, true) => !cast_to.is_signed(),
-
-        (_, _) => false,
+        // Don't lint float -> int casts as they have different semantics
+        _ => false,
     }
 }
 

tests/ui/cast.rs

@@ -47,7 +47,6 @@ fn main() {
 
     1f32 as u32;
     //~^ cast_possible_truncation
-    //~| cast_sign_loss
 
     1f64 as f32;
     //~^ cast_possible_truncation
@@ -63,12 +62,10 @@ fn main() {
 
     1f64 as usize;
     //~^ cast_possible_truncation
-    //~| cast_sign_loss
 
     1f32 as u32 as u16;
     //~^ cast_possible_truncation
     //~| cast_possible_truncation
-    //~| cast_sign_loss
 
     {
         let _x: i8 = 1i32 as _;
@@ -82,7 +79,6 @@ fn main() {
 
         1f32 as u8;
         //~^ cast_possible_truncation
-        //~| cast_sign_loss
     }
     // Test clippy::cast_possible_wrap
     1u8 as i8;