Fix N+1 query on convention/signup_rounds in the My Schedule widget

[nbudin]

Summary

Fixes INTERCODE-17E, a conventions + signup_rounds N+1 that was explicitly left open when the SignupCountPresenter N+1s were fixed in #11880 -- it was a different, unfixed code path even then.

Investigation

A scheduled Sentry check-in proposed that this was ConventionDrop#products/#available_products/#ticket_types calling .convention on each result without a preload (product.convention.timezone in ProductDrop/TicketTypeDrop). I wrote a test for that hypothesis first, per usual, and it didn't hold up: those methods all load via has_many associations off the single already-loaded convention, so Rails' automatic inverse-of association caching means .convention on each result is already free -- with or without an explicit .includes. Confirmed this and did not carry that dead-end change forward.

The actual site, found by re-checking the code that {% withdraw_user_signup_button %} runs (rendered once per signup in the same "My Schedule" widget already fixed once before for the bucket/requested_bucket N+1 in #11882):

signup_rounds = signup.event.convention.signup_rounds

UserConProfileDrop#signups already preloads event: :event_category, but never the event's convention -- so each signup's event is cached, but .convention on it isn't, and .signup_rounds on that freshly-loaded (per-signup) Convention instance re-queries too. Added convention: :signup_rounds to the existing event: preload.

Test plan

• New regression test in test/liquid_drops/user_con_profile_drop_test.rb, mirroring exactly what the Liquid tag does (signup.event.convention.signup_rounds) -- verified to fail without the fix (10 queries for 5 signups) and pass with it (≤2)
• Full test/liquid_drops/ suite (41 tests)
• rubocop / stree clean

🤖 Generated with Claude Code

[github-actions]

Code Coverage Report: Only Changed Files listed

Package	Base Coverage	New Coverage	Difference
app/javascript/setupI18Next.ts	🟠 65.52%	🟠 72.41%	🟢 6.89%
Overall Coverage	🟢 55.99%	🟢 56%	🟢 0.01%

Minimum allowed coverage is 0%, this run produced 56%