Fix regression in implicit casting from integer -> string

Modelling AML strings in Rust is tricky; conversion from integers with
`String::from_utf8_lossy` keeps trailing null bytes - this is likely fine
in C because of how string comparisons would work, but trips up comparison
in Rust's `String` (and `CString` cannot have multiple trailing null bytes).
This fixes a regression in uACPI's implicit_case_semantics by removing
trailing null bytes when doing an integer -> string conversion.

It may be that the correct place to fix this holistically is rather in the
object comparsion, but this seems like a reasonable step given the current
failing case.

Author: IsaacWoods

src/aml/object.rs

@@ -316,7 +316,7 @@ impl Object {
                 *value = u64::from_le_bytes(bytes);
             }
             Object::String(value) => {
-                *value = String::from_utf8_lossy(&new_bytes).to_string();
+                *value = String::from_utf8_lossy(&new_bytes).split('\0').next().unwrap().to_string();
             }
             Object::Buffer(value) => {
                 *value = new_bytes.to_vec();