Drop bulk_import_exports.organization_id column

What does this MR do and why?

Part 2 of 3 for #604997 (closed). Part 1 (!245994 (merged)) stopped the app from writing organization_id on bulk_import_exports and dropped the trigger, FK, and index. The model already ignores the column.

Now that part is deployed, so this drops the physical column. down restores it as a nullable bigint. Also clears out the now-stale sharding_key_spec and schema_spec exemptions since the column is gone.

Part 3 (#606168) removes the ignore_column.

Data deletion

Dropping the organization_id column on bulk_import_exports.

1. Recovery: down re-adds the column as a nullable bigint. The old values aren't restored, but they were only ever derived from the row's parent project or group (that's what the part 1 trigger did), so they can be re-derived if we ever need them. Nothing reads the column today.
2. Records affected: about 7.6M rows have the column, but it's been unused since part 1 shipped. The app stopped writing it, the model ignores it, and nothing reads it.
3. User experience impact: none. The column was never surfaced to users or read by the app.

References

• #604997 (closed), #606167 (closed)
• !245994 (merged) (part 1)

Related to #606167 (closed)