[lesnik512]

"Left behind by these picks" listed every relation of a shipped issue. A line like SHOP-35 blocks SHOP-40 → worker: no tag yet means SHOP-40 is waiting on SHOP-35, which is no reason to hold the release. Lines like that would bury the real warnings.

Changes:

• The list now follows only relations the shipped issue depends on: an inward link of Jira's Blocks type ("is blocked by"), and its subtasks. blocks, relates to, parents and any other link type are left out. They still show when you hover over the issue key in the Issues list.
• The match uses the link type name Blocks, Jira's default. An instance that renamed it or uses its own dependency type won't get warnings for it. If someone needs that, it could become a setting.
• Russian wording, reviewed and fixed:
  • Не войдёт с этими тегами: becomes Не войдёт в релиз с выбранными тегами:. The old version didn't say what wouldn't make it in.
  • родитель becomes входит в. "SHOP-41 родитель SHOP-36" read as if SHOP-41 were the parent.
  • подзадача becomes включает подзадачу, which had the same backwards reading.
  • нужен {tag} becomes нужен тег {tag}.
• The demo gets SHOP-35 blocks SHOP-40 (must not show) and SHOP-46, a To Do subtask of SHOP-36 (must show).
• docs/page.md describes which relations count.

Tests: the hand-pick test now expects the subtask line and no blocks line. It failed before the change by showing SHOP-35 blocks SHOP-40 → worker: no tag yet. A new test switches to Russian and checks the heading and two lines. The heading assertion has # noqa: RUF001, because ruff reads "Не" and "с" as look-alike Latin letters. just lint-ci, just test-ci and just test-smoke pass.