fix issue where a text block is extracted multiple times by different patterns

[fqrious]

When running pattern extractors, each extractor is run separately over the document and because of this a url and a domain can be extracted from a single text block. for example

host_url = "https://example.org/file.html"

extracts both

{
    "type": "url",
    "spec_version": "2.1",
    "id": "url--de2cb18e-af1f-59d9-ada1-8b9c89661257",
    "value": "https://example.org/file.html"
}
 

AND

{
    "type": "domain-name",
    "spec_version": "2.1",
    "id": "domain-name--3500c14b-9bde-5393-b4e5-b260aea7e058",
    "value": "example.org"
}

by using the start_index, we can skip all extractions that overlap