Editorial: Use 'list-concatenation' more

Specifically, use it in `Runtime Semantics: LabelledEvaluation`.

It's unusual to modify the value of an SDO parameter.
Certainly, other SDOs with a _labelSet_ parameter
are careful to copy it before appending anything.
(ContainsDuplicateLabels, ContainsUndefinedBreakTarget, ContainsUndefinedContinueTarget)

I'm pretty sure that the modification here doesn't matter
(because in the few places that LabelledEvaluation is invoked,
nothing is done with the _labelSet_ arg afterward).

But it's easier to understand (and more consistent)
if we leave _labelSet_ unmodified.
So I've used list-concatenation.

Author: jmdyck

spec.html

@@ -19663,8 +19663,8 @@ <h1>Runtime Semantics: LabelledEvaluation</h1>
       <emu-grammar>LabelledStatement : LabelIdentifier `:` LabelledItem</emu-grammar>
       <emu-alg>
         1. Let _label_ be the StringValue of |LabelIdentifier|.
-        1. Append _label_ as an element of _labelSet_.
-        1. Let _stmtResult_ be LabelledEvaluation of |LabelledItem| with argument _labelSet_.
+        1. Let _newLabelSet_ be the list-concatenation of _labelSet_ and &laquo; _label_ &raquo;.
+        1. Let _stmtResult_ be LabelledEvaluation of |LabelledItem| with argument _newLabelSet_.
         1. If _stmtResult_.[[Type]] is ~break~ and SameValue(_stmtResult_.[[Target]], _label_) is *true*, then
           1. Set _stmtResult_ to NormalCompletion(_stmtResult_.[[Value]]).
         1. Return Completion(_stmtResult_).