RegexpQueryNode inconsistent treatment of begin and end parameters [LUCENE-6498]

RegexpQueryNode has some questionable treatment of the begin and end parameters:

```Java
public RegexpQueryNode(CharSequence field, CharSequence text, int begin,
int end) {
this.field = field;
this.text = text.subSequence(begin, end);
}
```

when contrasted with FieldQueryNode:

```Java
public FieldQueryNode(CharSequence field, CharSequence text, int begin,
int end) {
this.field = field;
this.text = text;
this.begin = begin;
this.end = end;
this.setLeaf(true);
}
```

I am now not entirely sure which of the two is correct...

---
Migrated from [LUCENE-6498](https://issues.apache.org/jira/browse/LUCENE-6498) by Trejkaz

Start by comparing the RegexpQueryNode and FieldQueryNode constructors shown in the issue, then inspect how their begin and end parameters are used elsewhere. Determine the intended parameter semantics and verify the behavior with the relevant query-node tests, if present; the issue is done when the treatment is resolved consistently and covered by a regression test.

Written by the indexing model from the issue text.