Collect news text and image references without file attachments

Skip unsupported document formats and retire attachment work from active gap reporting. Preserve HTTP failures, normalize dynamic page content, and keep legacy bill queues resumable.

web/tests/crawler.test.ts

@@ -62,6 +62,8 @@ test("gap details retain reasons, exhausted state and legacy records while colle
   assert.equal(gaps[0].nextAt, new Date(1793853098 * 1000).toISOString())
   assert.equal(gaps[1].url, "https://www.justice.gov/release")
   assert.match(gaps[1].error, /图片/)
+  assert.equal(gaps[1].ignored, true)
+  assert.equal(gaps[0].ignored, false)
   assert.equal(gaps[1].exhausted, false)
   assert.equal(gaps[2].exhausted, true)
   assert.equal(gaps[2].url, "https://www.congress.gov/bill/119th-congress/house-bill/1")