Add an option to specify only target and causal dependency outlier metrics during inference for Do-Why

[garvitj742-code]

Is your feature request related to a problem? Please describe.
Add an option to specify only target and causal dependency outlier metrics during inference for Di-Why

Describe the solution you'd like
We want a do why inference function that only takes outlier metrics of causal dependencies and not all outlier metrics during inference. This will improve the overall time/latency and reduce extra overhead on clients.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.