refactor: use MigratorTraitSelf to eliminate global migration state - #2753

Please address the comments from this code review:

## Individual Comments

### Comment 1
<location path="migration/src/main.rs" line_range="7" />
<code_context>
 #[tokio::main]
 async fn main() {
-    cli::run_cli(migration::Migrator).await;
+    let storage = StorageConfig::parse()
+        .into_storage(false)
+        .await
</code_context>
<issue_to_address>
**Migration commands are rejected**

When the migration binary is invoked with a migration subcommand or its flags, `StorageConfig::parse()` parses the process arguments before `cli::run_cli` receives them, so migration commands such as `up` or `status` are rejected as unknown storage arguments and the migration CLI exits.

Parse storage settings without consuming the migration CLI arguments, or combine storage and migration arguments in one parser.
</issue_to_address>

### Comment 2
<location path="trustd/src/db.rs" line_range="15" />
<code_context>
     pub(crate) command: Command,
     #[command(flatten)]
     pub(crate) database: Database,
+    /// Location of the storage
</code_context>
<issue_to_address>
**Data command flags are rejected**

When a caller supplies storage or migration-option flags after `trustify db data`, clap defines `StorageConfig` and `Options` on `Run` rather than `Data`, so previously accepted flags after `data` are rejected as unknown arguments and the requested data migration does not run.

Preserve the flattened storage and migration options on `Data`, or make them global so existing command forms remain accepted.

Also at `trustd/src/db.rs:16-18`.
</issue_to_address>

### Comment 3
<location path="migration/src/main.rs" line_range="11" />
<code_context>
+        .into_storage(false)
+        .await
+        .expect("failed to initialize storage");
+    cli::run_cli(migration::Migrator::new(storage, ())).await;
 }
</code_context>
<issue_to_address>
**Configured migration options are ignored**

When a `MIGRATION_DATA_*` variable is set for the standalone migration binary, `Migrator::new(storage, ())` uses default `Options`, so configured skip and runner settings are ignored and data migrations run with defaults, including migrations the operator intended to skip.

Construct `Options` from the environment and pass it to the migrator.

Also at `server/src/profile/api.rs:465-466`, `xtask/src/dataset.rs:131`.
</issue_to_address>