Schema migration - tracking issue

implements the happy path of schema migration. do not use this feature yet.

Here a list of remaining work to get schema migration working properly:

• move all the migration-related stuff from the meta_store to the schema module; let the scheduler use its own connection to the meta store. (#1131)
• migration-specific result builder (#1146 )
• perform error handling, check migration success status, use custom QueryResultHandler to collect results (#1146)
• ensure that the temp table enqueued_tasks is updated corectly after a task is popped when a tasks status changed (maybe use a trigger?) (#1126)
• handle abort; first when a dry run failed, then if a user abort the migration before the dry-run is complete (#1163)
• prevent a job from being enqueued if there is still a pending job for that namespaces (#1131)
• prevent writes while there is are pending jobs in a database (https://github.com/tursodatabase/libsql/pull/1139)
• backup database before starting migration (#1140)
• backup meta store after enqueue job (#1144)
• track version schema
• go through all todos that I left in the code
• when performing a task, and that task was already performed, report the status, and don't re-perform the task (#1136)
• When a new db is created with a shared schema, bootstrap schema with current schema version. (#1154)
• prevent db creation while schema has pending migration it (#1137)
• admin API collect status of job (#1134)
• shared schema options should only be settable at creation (not possible through config update)
• ensure migration doesn't contain any transaction statements (#1149)
• deadlock in ns_loaded_with_pending_tasks_writes_is_blocked (#1173)
• prevent new migrations if a namespace creation is under way. (i.e. a new namespace creation should not miss fresh migration) (#1183)
• test schema delete

Start by reviewing the remaining unchecked items: the TODOs in the code and the missing schema-delete test. Trace the existing schema-migration work and linked issues to identify the relevant entry points and test locations. Done means the outstanding work is completed and schema deletion is covered by a test.