Implement selective template parsing for efficient wikitext parsing

[txetxedeletxe]

As of today, WikiText template parsing works like follows:

1. All Wikitext templates (even nested instances) are identified and extracted.
2. Templates are further parsed into template name and parameters.
3. Templates are filtered by name.
4. Paramaters on the filtered templates are further filtered.

Step 1 is very time expensive with complex and long documents, so it needs to be combined with step 3 to lower the amount of work done.

[txetxedeletxe]

Implemented the enhancement together with some additional improvements in commit: 0ac9a536115bf3fdd21d54ef66471ba953aee24a