Refactor schema scanners

[murphyatwork]

Refactoring request

Is your refactoring request related to a problem? Please describe.

SchemaScanner is the implementation at BE for the system tables in information_schema, it basically send an RPC request to FE to retrieve related information, then build the RPC result into a Chunk. But since each table can have different columns and individual rpc requests, the code is pretty verbose.

A good way to refactor these code is like be/src/exec/schema_scanner/schema_tasks_scanner.cpp:

• return an DatumArray as a row of chunk
• it doesn't need to handle the schema/slot anymore
• in this way it's more extensible to add a column in the table

DatumArray SchemaTasksScanner::_build_row() {
    auto& task = _task_result.tasks.at(_task_index++);
    if (!task.__isset.catalog) {
        // Compatible for upgrades
        task.catalog = "default_catalog";
    }
    Datum expire_time = task.__isset.expire_time && task.expire_time > 0
                                ? TimestampValue::create_from_unixtime(task.expire_time, _runtime_state->timezone_obj())
                                : kNullDatum;
    Datum create_time = task.__isset.create_time && task.create_time > 0
                                ? TimestampValue::create_from_unixtime(task.create_time, _runtime_state->timezone_obj())
                                : kNullDatum;

    return {Slice(task.task_name),  create_time, Slice(task.schedule),   Slice(task.catalog), Slice(task.database),
            Slice(task.definition), expire_time, Slice(task.properties), Slice(task.creator)};
}

Since there're many SchemaScanner in the codebase, only few of them have been refactored. we'd like to refactor all of them in this way be/src/exec/schema_scanner/

Describe the solution you'd like

Describe alternatives you've considered

Additional context