RFE: RR0150, RR0151 (Replace while with ...): Handle 'continue'

Roslynator Refactorings 2017 up to and including 2.1.1

When refactoring, and flipping conditional code around, you can sometimes end up with something like

```
int i = 0;
while (i cond1 smth)
{
if (cond2) { /* ... */ }
++i;
continue; // used to be inside "if (!cond2)
}
```

It would be nice if these refactorings noticed the `continue` is a NOP, allowing these two refactorings to run, and remove the (now spurious) `continue`.

Coming to think of it, that could/should probably be turned into an RR of itself - removal of pointless/"dead" code.

Additionally, it could be nice when converting to `for` to allow (though probably not enforce) the loop condition-variable initialization to be placed as such, converting the previous into

```
for (int i = 0; i cond1 smth; ++i)
{
if (cond2) { /* ... */ }
}
```

or perhaps that could be turned into an RR or two of itself, moving the for-loop variable initialization in/out of the for-initialization-expression. (just a brainstorming idea)

While I'm not the least familiar with the Roslynator code, a cursory glance suggest https://github.com/JosefPihrt/Roslynator/blob/d402258408dc8e17d737417e1c268f91be7b5d49/src/Refactorings/CSharp/Refactorings/ReplaceWhileWithForRefactoring.cs#L81-L82

could be the source of not allowing the refactoring? Of course, that's just the detection part...

Start with Refactorings/CSharp/Refactorings/ReplaceWhileWithForRefactoring.cs around lines 81-82, then trace the detection and transformation for RR0150 and RR0151. Determine how a loop containing continue should be handled and whether moving initialization into the for expression belongs in this change; done means the requested refactoring behavior is defined and implemented without the spurious continue.

Written by the indexing model from the issue text.