Allow to override the final table name in a database destination

[pulsar256]

Feature description

Currently the table names generated in a database destination (tested with ClickHouse) will be a function of dataset_name and dataset_table_seperator. If we use default values for those configuration options, most likely one will end with a table name generated as

{pipeline_name}_dataset__{resource_name}

This feature request is about adding an option to override the generation of the table name entirely bypassing any internal templating.

Are you a dlt user?

Adopting

Use case

Drop in replacement for other data movement tools which do not expect the downstream pipeline to adapt to naming conventions but give the user full control over destination table names.

Proposed solution

perhaps expose a table naming template string user can specify in the pipeline definition, i.E.:

    pipeline = dlt.pipeline(
        pipeline_name="test",
        destination="clickhouse",
        dataset_name_template="{resource_name}"
    )

[rudolfix]

@pulsar256 we do it to simulate schemas on destinations that do not support it. typically we let users to pass empty dataset name to pipeline to get rid of the naming layout but I see that this is disabled for clickhouse. we can enable it in the next release, seems like not much work