short_circuit_statement: handle macros and parenthesis better

[samueltardieu]

• The lint no longer triggers if one of the operands in the boolean expression comes from a macro expansion.
• Parenthesis are now removed inside the generated block if they are no longer necessary.
• Error markers have been added.

changelog: [short_circuit_statement]: better handling of macros and better looking suggestions

[rustbot]

r? @y21

rustbot has assigned @y21.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[y21]

The lint no longer triggers if one of the operands in the boolean expression comes from a macro expansion.

Isn't it fine to lint something like m1!() && m2!();? This could/should still be written as an if expression. I guess you want to rule out the mac!(f and g); case in the tests, but isn't the issue there moreso that the binary expression and semicolon statement are in different SyntaxContexts? So I think we would just want to check that the binary expression and statement aren't from an expansion, rather than the binop operands.

[samueltardieu]

The lint no longer triggers if one of the operands in the boolean expression comes from a macro expansion.

Isn't it fine to lint something like m1!() && m2!();? This could/should still be written as an if expression. I guess you want to rule out the mac!(f and g); case in the tests, but isn't the issue there moreso that the binary expression and semicolon statement are in different SyntaxContexts? So I think we would just want to check that the binary expression and statement aren't from an expansion, rather than the binop operands.

Done, and indeed this is better.