Overflow the last rhs of a binary expression

[topecongiro]

EDIT: this is no longer true.

---

This PR allows the last rhs of a binary expression to be overflowed if it is a chain.

We do not want to overflow the rhs of a binary expression that is inside the other binary expression. For example, we do not want to do these:

-        self.inner_as_ref().contains('\n')
-            || self.pre_comment
-                .as_ref()
-                .map_or(false, |s| s.contains('\n'))
-            || self.post_comment
-                .as_ref()
-                .map_or(false, |s| s.contains('\n'))
+        self.inner_as_ref().contains('\n') || self.pre_comment
+            .as_ref()
+            .map_or(false, |s| s.contains('\n')) || self.post_comment
+            .as_ref()
+            .map_or(false, |s| s.contains('\n'))

-                if !forbid_same_line⏎
-                    && (is_block⏎
-                        || (!body_str.contains('\n') && body_str.len() <= body_shape.width)) =>⏎
+                if !forbid_same_line && (is_block⏎
+                    || (!body_str.contains('\n') && body_str.len() <= body_shape.width)) =>⏎

Closes #2493.

[nrc]

Whilst I agree that #2493 looks bad, I think that the changes in the forth commit here are sub-optimal since they make things harder to read, IMO, because at a glance it's easy to see the chain, but miss the binary operators. I wonder if we should only do this change if the regular formatting would 'orphan' a binop as in #2493? We do something similar for chains - keeping the second element on the same line as the first if it would be an 'orphan' on its own line.

[topecongiro]

I think that the changes in the forth commit here are sub-optimal since they make things harder to read, IMO, because at a glance it's easy to see the chain, but miss the binary operators.

Actually I initially liked the change in the fourth commit, but after looking it back, they look bad.

I will update this PR so that we put the rhs of a binary expression only when the lhs is shorther than or equal to the tab width.

[topecongiro]

I especially dislike this change 👎

[topecongiro]

I have updated this PR so that we put both the lhs and the rhs on the same line when the width of the lhs is equal to or shorter than the tab width.

[nrc]

Thanks!