Make grammar for switch cases unambiguous (or make it clear that it is)

[lrhn]

The grammar for a switch case is (effectively):
  label? ('case' expression ':')+ statement*
Since statements* can be empty:
  case 1:
  case 2:
    break;
can be parsed as both one case with two expressions or two cases with one expression each, and no statements on the first one.

The last parsing is most likely wrong (implementations, but not the spec, requires a case to not reach the end without an explicit abort, and I am being told that it's intended to become specified too).

The problem is likely that +/* repetitions don't have any precedence or associativity associated with them. If they were specified as preferring more matches over fewer, the above would have a preferred parsing.
(Or maybe I'm just missing the place where this is already written, in which case it bears repeating near the switch grammar).

Alternatively, change statement* to statement+. It only changes something for the very last case of a switch, which then can't end as "case 4:}".
If we require explicit aborts from all cases.
  

[gbracha]

Set owner to @gbracha.
Added Accepted label.

[gbracha]

Added this to the M1 milestone.

[gbracha]

Addressed in 0.09.

---

Added Done label.