- Update CLI to accept multiple -e flags for entity columns
- Update mapper and codegen for multiple entities
- Update documentation with examples and usage

This extends the dbt integration to support FeatureViews with
multiple entities, enabling use cases like transaction features
keyed by both user_id and merchant_id.

Fixes feast-dev#5872

Signed-off-by: yassinnouh21 <[email protected]>

When creating entities from dbt models with integer columns, entities were
being created with default ValueType.STRING, causing validation errors:
"Entity X has type ValueType.STRING, which does not match the inferred type Int64"

Solution: Added mapping dict and helper function for clean type conversion:
- FEAST_TYPE_TO_VALUE_TYPE: Maps FeastType to ValueType
- feast_type_to_value_type(): Helper function for conversion
- _infer_entity_value_type(): Method in DbtToFeastMapper class

This replaces verbose if-else chains with a clean dictionary lookup pattern.

Also corrects schema generation to include entity columns, as FeatureView.__init__
expects to extract entity columns from the schema itself (lines 216-234 in
feature_view.py).

Changes:
- feast/cli/dbt_import.py: Use mapper's _infer_entity_value_type() method
- feast/dbt/mapper.py: Add type mapping dict and helper method

Signed-off-by: yassinnouh21 <[email protected]>

Updated test_mapper.py to use new parameter names:
- entity_column -> entity_columns (accepts both str and List[str])
- result["entity"] -> result["entities"] (now returns list)

Tests now correctly check for list of entities returned by
create_all_from_model().

Signed-off-by: yassinnouh21 <[email protected]>