Improve reference location handling by specifying the correct start range

see https://github.com/camel-tooling/camel-language-server/pull/192 comment of @apupier :

- the current end character is the same than the start one. Was it a problem with previous implementation searching in same file? if yes, can you create an issue to improve that? If not, might worth adding a check on position end in test and improve it.
example of result:
```
[Trace - 3:23:43 PM] Received response 'textDocument/references - (25)' in 15ms.
Result: [
{
"uri": "file:///c%3A/Users/Aurelien%20Pupier/Desktop/camel4.xml",
"range": {
"start": {
"line": 5,
"character": 40
},
"end": {
"line": 5,
"character": 40
}
}
}
]
```

Start with the linked pull request #192 and the textDocument/references response shown in this issue. Trace how reference ranges are produced, then inspect the relevant test or add a check for the end position. Done means returned ranges have the correct end character instead of duplicating the start character.

Written by the indexing model from the issue text.