Option to migrate SQL Server tables to PostgreSQL

[labkey-adam]

Rationale

We want to provide administrators the ability to easily migrate a SQL Server LabKey SDMS deployment to PostgreSQL. Changes in this PR provide an option to create empty PostgreSQL schemas and migrate all module & provisioned table data into them from a SQL Server deployment.

Specification: https://docs.google.com/document/d/1oY3Mhnusa17OsC3NDj3WR5gS5YlynUP5zzc2tBht_D4/edit?tab=t.0#heading=h.4rg4h5vuek0

This is a significant step forward, but it's still a work in progress. What's missing:

• Testing with non-trivial SQL Server databases
• Testing with large amounts of data

There should be no functional changes in the normal case. The new code is invoked only when the server is started with specific command-line arguments.

Changes

If a PostgreSQL database is being bootstrapped and a specific argument is added to the command line, the server will:

• Bootstrap all PostgreSQL schemas without inserting any* data into them
• Flag all PostgreSQL tables that still contain rows and sequences set to a value other than one, plus a few other sanity checks
• Connect to the specified SQL Server database as an external data source and enumerate all module tables, sorted to accommodate foreign key constraints
• Copy data from every table into the just-created empty PostgreSQL tables
• Reset the PostgreSQL sequences to ensure that future keys produced by SERIAL columns don't conflict with the copied-in data
• Create all the provisioned tables based on the rows in exp.DomainDescriptor
• Repeat the data migration and sequence updates for the provisioned schemas/tables

*Some details:

• The root and shared containers are actually created for bootstrap purposes (many code paths rely on their presence) and then deleted before populating the tables
• The core Modules, SqlScripts, and UpgradeSteps tables are populated by the PostgreSQL bootstrap process and not cleared. The data in these SQL Server tables is not migrated.
• The migrationDataSource argument is used to specify the SQL Server data source to migrate. For example: -DmigrationDataSource=ssDataSource
• -DemptySchemas=true can be used to test the empty schema creation and verification without attempting an actual migration
• While not typically needed, a few modules register schema-specific MigrationHandler implementations to customize behavior (e.g., accommodate cyclical FKs or adjust the tables to migrate)

Tasks 📍

• Option to bootstrap schemas without any data
• Verify empty schemas and sequences
• Copy data from SQL Server tables
• Adjust sequences to accommodate copied data
• Fix TableSorter handing for self-referencing foreign keys
• Special case inserts for cyclical FKs in wiki (Pages/PageVersions), prot (Organisms/Identifiers), exp (ExperimentRun/ProtocolApplication)
• Expand testing to all modules and ensure empty lookup tables throughout
• Drop problematic, unused Deleted columns from prot schema
• Update spec
• Create provisioned tables and migrate data
• Code review @labkey-jeckels
• Manual Testing @labkey-danield 📍
• Needs Automation - Not at this point. Eventually, we'll need some automated testing.