#14 · Multi-tenant config: load tenant-specific env from `tenants/<id>/.env`

Context

Forks of the bot can't easily run multiple Signal accounts on one host. A multi-tenant layout (tenants/<tenant_id>/.env, tenants/<tenant_id>/state/) would let one container serve N communities with isolated state, sharing only the binary. This is foundational for the fork-friendly direction in selfhost/.

Acceptance criteria

• [ ] Bot reads TENANT_ID env var; if set, loads tenants/$TENANT_ID/.env instead of root .env
• [ ] All file paths (signal-cli state, file index, logs) namespace under tenants/<id>/
• [ ] Database connection string uses DB_NAME=signalbot_<tenant_id> if multi-tenant mode active
• [ ] Default behavior unchanged when TENANT_ID unset (single-tenant compatibility)
• [ ] docker compose up with TENANT_ID=foo example documented in README
• [ ] Tests: at least one integration test runs in multi-tenant mode

Technical hints

• Centralize env loading in apps/signal-bot/src/src/utils/config.ts (probably already exists — extend it)
• Update compose.yml to honor TENANT_ID build arg / env
• PostgreSQL: per-tenant database, NOT per-tenant schema (avoids cross-tenant leaks via shared sequences)
• signal-cli stores account state in ~/.local/share/signal-cli/ — must override per-tenant

--- *Submit a merge request that says Closes #<this issue iid> in the description. On merge to main, the GitLab webhook will auto-award your bounty + the standard merge_pr (40 pts) to your scoreboard account. Email on commit must match your Authentik email.*