fix: skip analyze/tag/extract_links for link-sourced stub items

Add _source=link guard to after-write fragments for analyze, auto_tag,
and extract_links. Stub items created by URL-following (extract_links
with create_targets=true) now skip expensive background processing.
Prevents wasteful work and blocks potential recursive crawling.

resolve_stubs already had this guard.

Author: hughpyle

keep/builtin_state_docs.py

@@ -324,19 +324,19 @@
 """,
         "analyze": """\
 rules:
-  - when: "!item.is_system_note"
+  - when: "!item.is_system_note && !(has(item.tags._source) && item.tags._source == 'link')"
     id: analyzed
     do: analyze
 """,
         "tag": """\
 rules:
-  - when: "!item.is_system_note && item.has_content"
+  - when: "!item.is_system_note && item.has_content && !(has(item.tags._source) && item.tags._source == 'link')"
     id: tagged
     do: auto_tag
 """,
         "links": """\
 rules:
-  - when: "!item.is_system_note && item.has_content && item.content_type == 'text/markdown'"
+  - when: "!item.is_system_note && item.has_content && item.content_type == 'text/markdown' && !(has(item.tags._source) && item.tags._source == 'link')"
     id: linked
     do: extract_links
     with: