DStyle: Constraints on declarations should have the same indentation level

[wilzbach]

I accidentally closed #4999 - the decision from this PR was:

It is a problem. Phobos is a messy mix of styles, which makes it gratuitously difficult to work with. Let's not argue this please - it's subjective so a decision needs be made. @wilzbach please rebase and let's pull. Thanks!
Expect future improvements to the style guide followed by automated enforcements.

I updated the replacement tool that looks at all Dscanner warnings and tries to fix them, so that a couple of previously unhandled edge case are handled. Moreover it now also updates the indentation of the following lines of the if constraint correctly (+4 of the if indentation level).

Steps to reproduce

1) Git clone the specific branch of the dscanner

git clone https://github.com/wilzbach/dscanner ../dscanner
git -C ../dscanner if_constraints_same_indent

2) Modify dscanner to emit the expected indent size

--- a/src/analysis/if_constraints_indent.d
+++ b/src/analysis/if_constraints_indent.d
@@ -101,8 +101,10 @@ private:
                                                        .until!(x => !x.isWhite)
                                                        .walkLength;
 
+         import std.conv : to;
                if (tokenIndent != lineIndent)
-                   addErrorMessage(constraint.expression.line, constraint.expression.column, KEY, MESSAGE);
+                 addErrorMessage(constraint.expression.line, constraint.expression.column, KEY,
+                         MESSAGE ~ ":" ~ tokenIndent.to!string);
        }
 }

3) Ignore other checks

Add the following to the .dscanner.ini:

allman_braces_check = "disabled"
trailing_whitespace_check = "disabled"
consecutive_empty_lines = "disabled"

4) Run dscanner and save all warnings

make -f posix.mak style > list

5) Let the replacement tool fix all warnings

rdmd --compiler=../dmd/src/dmd ../replacements/replacement.

[skl131313]

Shouldn't this line still be indented? The || is part of a sub condition.

(isForwardRange!ForwardRange || (isInputRange!ForwardRange && isInfinite!ForwardRange))

Putting it at the same level makes it confusing, with the && and || aligned.

[andralex]

I agree with @sprinkle131313 but let's leave that to the next pass.

[andralex]

Auto-merge toggled on

[andralex]

@wilzbach all good, rebase and babysit the autotester