[codex] Fix later-round scorecard generation

[coder13]

What changed

This PR was generated entirely by AI assistance.

It fixes later-round scorecard generation when a round already has competitor assignments but the WCIF does not yet have populated result scaffolding for that round.

The functional changes are:

• treat later rounds with existing assignments but empty results arrays as needing scorecards
• generate later-round scorecards from actual persons[].assignments when those assignments exist, instead of falling back to blank nameless cards
• add regression tests covering assigned later rounds with empty results

Why these code changes were needed

src/logic/activities.js

roundsMissingScorecards() now evaluates all rounds through a small helper instead of relying on roundsMissingResults().

I made that change because roundsMissingResults() was too narrow for this case. A later round can already have assignments and still have an empty results array. In that state, we still want the round to show up as scorecard-ready.

The new helper keeps the rule explicit:

• first rounds with no results still need scorecards
• later rounds with no results also need scorecards if they already have group assignments
• rounds with result rows but no attempts still need scorecards as before

src/logic/documents/scorecards.js

groupActivitiesWithCompetitors() now builds the group list before deciding whether it can derive competitors from results.

I made that change because the old flow only had one source of truth for later rounds: competitorsForRound(). That works when the WCIF already has the expected round/result scaffolding, but it fails when assignments exist before those result rows are populated.

The new flow adds a second path:

• gather the round's group activities first
• look directly at wcif.persons[].assignments for competitor assignments in each group
• if assignments exist, use them to build scorecards
• if assignments do not exist, keep the previous fallback behavior that generates nameless placeholder scorecards

That is why the diff in scorecards.js is larger than a simple condition change. This fix needed a real assignment-based path, not just a different boolean.

The direct assignment path also sorts competitors in a stable order. It prefers known round ordering when available, then station number, then name. I added that so printing stays consistent and readable even when round rankings are not available yet.

scorecards() is now exported so the regression test can exercise the actual scorecard-generation path directly.

Impact

• assigned later rounds show up as scorecard-ready
• scorecards for those rounds include competitor names and stations when assignments are present
• rounds with no known competitor assignments still fall back to nameless placeholder scorecards as before

Validation

• yarn test --watch=false --runInBand src/logic/tests/activities.test.js src/logic/tests/scorecards.test.js src/logic/tests/competitors.test.js src/components/App/App.test.js src/components/Header/Header.test.js src/components/CompetitionList/CompetitionList.test.js src/logic/tests/formatters.test.js