#289: stop lifetime as normal completion

[GHX5T-SOL]

Fixes #289.

Stops --lifetime expiration from being reported as a per-judge --timeout error when the lifetime timeout fires while a judge is running. The update now logs lifetime termination, still exports the factbase and churn file, and preserves the existing per-judge --timeout error path.

Validation:

• env PICKS=yes mise x ruby@3.3 -- bundle exec ruby -Itest -Ilib test/commands/test_update.rb --name '/lifetime|timeout|churn/' -> 6 tests, 17 assertions, 0 failures
• env PICKS=yes mise x ruby@3.3 -- bundle exec ruby -Itest -Ilib test/commands/test_update.rb -> 28 tests, 64 assertions, 0 failures
• mise x ruby@3.3 -- bundle exec rubocop lib/judges/commands/update.rb test/commands/test_update.rb -> 2 files inspected, no offenses
• git diff --check
• git diff --no-ext-diff | gitleaks stdin --no-banner --redact --exit-code 1 -> no leaks found

Limitations:

• mise x ruby@3.3 -- bundle exec rake test ran 137 tests and 423 assertions with 0 failures, but hit one local tidy -e .../temp/base.html error in test_print_to_html.
• Full bundle exec rubocop reports four existing symbol-conversion warnings in unrelated command files.

[yegor256]

@GHX5T-SOL merge conflict here

[yegor256]

Please fix the merge conflicts so this pull request can be merged.

[yegor256]

Please fix the merge conflicts so this pull request can be merged.

[edmoffo]

APPROVE. The fix in lib/judges/commands/update.rb at lines 75-82 drops the conditional re-raise so the lifetime Timeout flows into the ensure block unchanged, and the new branch at lines 258-268 inside one_judge distinguishes a true per-judge --timeout from a lifetime expiration that surfaces inside a judge run, which is exactly what #289 reported. test_terminates_on_lifetime and test_exports_churn_to_file_on_lifetime now assert the completion path instead of expecting a raised error, which matches the new contract. CI is green across rake on ubuntu, macos, and windows, plus actionlint, rubocop, reuse, and the rest. The branch is behind master, so rebasing before merge will produce a smaller integration than the current head-vs-master comparison suggests.

[0crat]

@edmoffo Thanks for the review! You've earned +10 points for this: +18 as a basis; -8 for absolutely no comments posted. Your running score is +1913; don't forget to check your Zerocracy account too).

[yegor256]

@GHX5T-SOL merge confict here