Implemented macro for compile time non empty string creation

[patskovn]

This is more convenient for constant strings in the codebase that are known to be non empty.

Proper tests for failures would require importing special libraries for macro tests. I do not expect this to be too so I assume it is better to keep this lib smaller and not to add unit tests for compilation failures. Thankfully doctests validate the behaviour, but to be sure I tested it manually via:

diff --git a/src/macros.rs b/src/macros.rs
index a1720fd..0b411cc 100644
--- a/src/macros.rs
+++ b/src/macros.rs
@@ -41,4 +41,9 @@ mod tests {
         let s = non_empty!("Test String");
         assert_eq!(s, NonEmptyString::try_from("Test String").unwrap());
     }
+
+    #[test]
+    fn test_non_empty_string_macro_empty_runtime() {
+        let _ = non_empty!("");
+    }
 }

And then

$ cargo t --features macro

error[E0080]: evaluation of constant value failed
  --> src/macros.rs:48:17
   |
48 |         let _ = non_empty!("");
   |                 ^^^^^^^^^^^^^^ the evaluated program panicked at 'String cannot be empty', src/macros.rs:48:17
   |
   = note: this error originates in the macro `$crate::panic::panic_2021` which comes from the expansion of the macro `non_empty` (in Nightly builds, run with -Z macro-backtrace for more info)

For more information about this error, try `rustc --explain E0080`.
error: could not compile `non-empty-string` (lib test) due to 1 previous error
warning: build failed, waiting for other jobs to finish...

[mhedgpeth]

+1 to this PR; I went looking for this functionality, copied in the macro, and this really helped me not make the crate to difficult to use when I had obviously non-empty static strings.

[MidasLamb]

Thanks for the work @patskovn ! I've branched off on your initial commit and merged it in through #20 (You've waited long enough for any feedback on this PR, didn't want to make you wait any longer just to revert your second commit).