Name the handle, and say at boot which badges have no emoji

[maersux]

Three things, and the middle one is why the first was needed.

The translator emoji. BADGE_EMOJI had eight entries and the api now sends nine. A row falls back to plain names unless every badge on it has an emoji, so the first person granted the badge would have lost their whole row — donator emoji included. Nobody held it yet, so nothing was broken; it was waiting.

The test that should have caught it was the problem. It read "carries an id for every badge, so a row never falls back for one missing entry" and checked against a hardcoded list of eight — a third copy of a list the api owns. Map, list and test agreed and were wrong together. A test cannot ask production, so the completeness check moved to boot: the bot asks /v1/badges once and names what it cannot draw. Verified against the live api — it printed no emoji for translator before the id went in, and badge emoji cover all 9 badges after.

The join and boost lines name the handle. Kyelin / just joined the server! becomes Kyelin / **kyelin** just joined the server! — the author line keeps the display name, the line under it says who that is, which is the part that survives a rename.

The description is markdown and the author line is not, which this repo already had a test for. A discord username may hold _, so some_user_name would have arrived with its middle in italics and the underscores eaten. Escaped, with a test.

64 tests green, prettier clean.

Note for the deploy: DISCORD_BADGE_ROLES on the server needs ,translator:1542117530553032755 appended for the badge to grant the role. No code change — /v1/internal/badges/discord is generic and already returns translator holders.