fix(team-randomizer): add safety improvements

[lbzepoqo]

Summary

Several safety and UX issues in TeamRandomizer that affect production servers:

• Disabled by default — changed defaultEnabled from true to false. This is a destructive, irreversible-during-round operation; it shouldn't be on without explicit opt-in.
• Re-entrancy guard — concurrent !randomize calls from multiple admins would run simultaneously and clobber each other. Added a this.randomizing flag with try/finally to ensure it always resets even on error.
• Error handling per player — the original loop had no try/catch; a single failed switchTeam RCON call would abort mid-shuffle, leaving teams in a partially-randomized state. Now each switch is caught individually, logs the failure, and continues.
• Player broadcast — players were silently moved to different teams with no explanation. Added an AdminBroadcast before switching starts.
• Admin feedback — the admin who ran the command got no confirmation. Added a warn on completion with the count of players switched.
• Shuffle cleanup — removed the unnecessary temporaryValue temp variable; replaced with destructuring swap.

Test plan

• Run !randomize in admin chat — confirm broadcast fires, teams are shuffled, admin receives completion message
• Run !randomize twice simultaneously — confirm second call is rejected with "already in progress" message
• Verify plugin is opt-in by default (not enabled in generated config)

[fantinodavide]

@lbzepoqo the plugin file itself looks good, but the pre-commit hook didn't run, which is a mandatory step for changes like this one.

[lbzepoqo]

Pre-commit hook steps have now been applied: lint (ESLint + Prettier) reformatted team-randomizer.js, and yarn build-all regenerated README.md and config.json. Sorry for the oversight!