collapsing switch cases with same bodies, when not lined in a row

[elad-yosifon]

This is an additional optimization that leverages changes from #1044 @jridgewell

Current code does not fully optimize the following code:

 switch (_var) {
    case 1: return true;
    case 2: return true;
    case 3: return false;
    case 4: return true;
    case 5: return false;
    case 6: return false;
    case 7: return true;
    case 8: return true;
}

The current output returns something like this:

 switch (_var) {
    case 1: case 2: 
      return true;
    case 3: 
      return false;
    case 4:
      return true;
    case 5: case 6: 
      return false;
    case 7: case 8: 
      return true;
}

The expected output in this case should be:

switch (_var) {
    case 1: case 2: case 4: case 7: case 8: 
      return true;
    case 3: case 5: case 6: 
      return false;
}

---

Adding a default case can benefit the optimization even more:

 switch (_var) {
    case 1: return true;
    case 2: return true;
    case 3: return false;
    case 4: return true;
    case 5: return false;
    case 6: return false;
    case 7: return true;
    case 8: return true;
    default: return true;
}

The expected output here should be:

switch (_var) {
    default: 
      return true;
    case 3: case 5: case 6: 
      return false;
}

[fabiosantoscode]

Wondering if @jridgewell wants to give a final approval. It looks great to merge to me, especially those tests <3.

[jridgewell]

I'd already approved. 😉

[elad-yosifon]

@fabiosantoscode @jridgewell great, what's next?

[jridgewell]

@fabiosantoscode will be able to merge. For the next PR, we could work on getting inert_branch performing better (the test case in #1070 (comment) isn't eliminating as well as it could), or work on reordering in between side-effect branch expressions (#1070 (comment)).

[fabiosantoscode]

Hey! This is indeed good to merge! Thanks @elad-yosifon 🚀

[LongTengDao]

@fabiosantoscode maybe a option to disable this would be better? because runtime order may be optimized, and will slow if do this

[elad-yosifon]

@LongTengDao ATM this is buggy, so I think this should be disabled but not discarded.
I might be able to do some work on it this weekend..