Make sure (most) queries avoid table scans.

I've exempted inserts (they never scan in the first place), queries on `event_sequence` (at most one row), and the coalesce()s used for event replay (for now; these are obviously a performance risk area and need addressing).

Method:

```
find .sqlx -name 'query-*.json' -exec jq -r '"explain query plan " + .query + ";"' {} + > explain.sql
```

Then go query by query through the resulting file.

1 files changed, 12 insertions, 0 deletions

diff --git a/.sqlx/query-2f26bccedd1226764e49f00a94e43226ee6aebdd14d7be97a708d66263a30abf.json b/.sqlx/query-2f26bccedd1226764e49f00a94e43226ee6aebdd14d7be97a708d66263a30abf.json
new file mode 100644
index 0000000..686afb9
--- /dev/null
+++ b/.sqlx/query-2f26bccedd1226764e49f00a94e43226ee6aebdd14d7be97a708d66263a30abf.json
@@ -0,0 +1,12 @@
+{
+  "db_name": "SQLite",
+  "query": "\n                insert into message_deleted (id, deleted_at, deleted_sequence)\n                values ($1, $2, $3)\n            ",
+  "describe": {
+    "columns": [],
+    "parameters": {
+      "Right": 3
+    },
+    "nullable": []
+  },
+  "hash": "2f26bccedd1226764e49f00a94e43226ee6aebdd14d7be97a708d66263a30abf"
+}