Add support to programmatically provide TableField values to RecordStreams

This is to support use cases where table data comes from an outside source (files or services) or needs to be generated programmatically (vs pulled in directly from the configured appDb with SQL).

Design plan is as follows:

1. In the model XML, add option to replace the queryRef attribute with a javaClass value which directs WDK to instantiate and call a Java plugin class to provide table data instead of an SQL query.
2. This plugin will take paged, sorted ID SQL and provide a Stream/Iterator of String[] containing the PK columns for the recordclass and the columns of the table (same columns table SQL provides).
3. RecordStream implementations must check to see what kind of table field provider this is (SQL vs Java) and branch there, providing rows either from the ResultSet or from the plugin (probably will create a plugin impl that takes SQL and call that).

Start with the model XML handling of queryRef and the RecordStream implementations that currently read table rows from ResultSet. Trace how paged, sorted ID SQL and table columns reach those streams, then determine the plugin entry point described in the issue. Done means Java-provided TableField values work alongside SQL-backed values for the same record and paging behavior.