feat(wordpress): only assign tags that are established

_resolve_wp_tag_ids created a WordPress tag for every keyword the rewriter
invented, up to 12 per post. That is where the 3.055 tags for 955 posts
came from - 1.683 of them used exactly once, 473 attached to no post at
all. The categories are stable now, but the tags would simply grow back.

Three changes, all on the write path:

A proposed tag has to appear for wordpress_new_tag_min_proposals (3)
different articles before it is created. Proposals are counted in the new
tag_proposals table, keyed on (name, article), so re-publishing an article
does not inflate its own count. Tags that already exist in WordPress are
assigned as before - the gate only guards creation.

Only the first wordpress_max_tags_per_post (5) tags reach WordPress. The
full list still feeds the category rules, which were validated against it.

The lookup fallback of reusing the first search hit is gone. It filed
"Camping" under the unrelated existing tag "Campingplatz" whenever the
exact tag was missing, which quietly produced wrong tags rather than none.

If the proposal bookkeeping fails, nothing is creatable that run: existing
tags still get assigned and the taxonomy stays put, rather than falling
back to creating everything.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

backend/app/repositories.py

@@ -853,3 +853,41 @@ def list_articles(limit: int = 100, status_filter: str | None = None) -> list[di
                 (safe_limit,),
             ).fetchall()
     return rows_to_dicts(rows)
+
+
+def record_tag_proposals(names: list[str], article_id: int | None) -> dict[str, int]:
+    """Record that these tags were proposed for one article.
+
+    Returns, per casefolded name, how many distinct articles have proposed it so
+    far - this one included. That count is what decides whether a tag is
+    established enough to be created in WordPress: a tag the rewriter invents
+    once for a single article should not become a permanent taxonomy entry.
+
+    Re-publishing an article does not inflate the count, because the row is
+    keyed on (name_key, article_id). Articles without an id (ad-hoc publishing)
+    all share the sentinel 0 and therefore count once in total.
+    """
+    counts: dict[str, int] = {}
+    if not names:
+        return counts
+    key_for_article = int(article_id or 0)
+    with get_conn() as conn:
+        for raw in names:
+            name = str(raw or "").strip()
+            if not name:
+                continue
+            name_key = name.casefold()
+            conn.execute(
+                """
+                INSERT INTO tag_proposals (name_key, name, article_id)
+                VALUES (?, ?, ?)
+                ON CONFLICT(name_key, article_id) DO NOTHING
+                """,
+                (name_key, name, key_for_article),
+            )
+            row = conn.execute(
+                "SELECT COUNT(*) AS n FROM tag_proposals WHERE name_key = ?",
+                (name_key,),
+            ).fetchone()
+            counts[name_key] = int(row["n"] if row else 0)
+    return counts