Coordinator crashes on :remove mutation for unknown index during initial app-state sync

[sh41]

Description

Syncd.Codec.mix_mutation/2 raises a RuntimeError when it receives a :remove mutation for an index that doesn't exist in the local index_value_map. This crashes the Coordinator GenServer on every fresh pairing during regular_low app-state sync.

Version: 0.1.0-alpha.7
Elixir: 1.19.5-otp-28
Erlang: 28.1

Error

** (RuntimeError) tried remove, but no previous op for index 22ih2vBBdIjulk01vd1PQi/qIdChj611M+cCvLN5I9E=
    (baileys_ex) lib/baileys_ex/syncd/codec.ex:170: BaileysEx.Syncd.Codec.mix_mutation/2
    (baileys_ex) lib/baileys_ex/syncd/codec.ex:326: BaileysEx.Syncd.Codec.decode_single_mutation/5

Steps to reproduce

1. Delete auth state directory (rm -rf tmp/baileys_auth)
2. Pair a fresh session
3. Post-pairing 515 restart → reconnect
4. Coordinator crashes during regular_low app-state resync

Analysis

The upstream JS Baileys has the same throw new Boom('tried remove, but no previous op') in mix():

• src/Utils/chat-utils.ts line 88

However, in JS the caller in chats.ts wraps decodePatches / decodeSyncdSnapshot in a try/catch (lines 559–620) that:

1. Catches the error
2. Logs it
3. Nulls the collection state (app-state-sync-version: { [name]: null })
4. Retries from scratch (up to MAX_SYNC_ATTEMPTS)

The BaileysEx Coordinator does not have equivalent error handling around the decode pipeline, so the raise propagates uncaught and kills the GenServer.

Possible fixes

Either:

• (A) Add try/rescue handling in the Coordinator's app-state resync (matching the JS try/catch + retry pattern), or
• (B) Change mix_mutation/2 to skip the no-op remove gracefully (return gen unchanged) — which is arguably more defensive since Map.delete on a missing key is harmless anyway

Option (A) more closely mirrors JS Baileys. Option (B) is simpler and prevents the error from occurring at all.