fix(evaluator): fix off-by-one in SpellingBeeEvaluator

[Copilot]

client_ids is stored in reverse order (last letter first). The current bloop is already verified against client_ids[0], so only N-1 prior bloops need to be fetched from global_bloops(). Using take(N) caused the comparison to silently fail whenever extra older bloops existed in history, because the iterator lengths would be unequal and eq() would return false.

The existing tests passed by coincidence — each had exactly N-1 prior bloops, so take(N) never grabbed an extra item.

Changes

• src/evaluator/spelling_bee.rs
  • take(self.client_ids.len()) → take(self.client_ids.len() - 1)
  • Adds regression test matches_correct_sequence_with_prior_bloops: correct sequence with an extra older bloop in history that would have triggered the bug

// Before
.take(self.client_ids.len())           // takes N, but skip(1) only has N-1 → fails with any prior history
.map(|bloop| &bloop.client_id);

last_client_ids.eq(self.client_ids.iter().skip(1))

// After
.take(self.client_ids.len() - 1)       // correctly takes only the N-1 remaining letters
.map(|bloop| &bloop.client_id);

last_client_ids.eq(self.client_ids.iter().skip(1))

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[github-actions]

🎉 This PR is included in version 1.10.8 🎉

The release is available on:

• v1.10.8
• GitHub release

Your semantic-release bot 📦🚀