Support multiple treatments in Causal ML

**Is your feature request related to a problem? Please describe.**
My causal ML problem has multiple treatment variables. I am trying it with EconML and it works but it needs too much memory - it doesn't scale up to big data. But it looks like SynapseML causal ml only supports a single treatment. Or am I misunderstanding it.

**Describe the solution you'd like**
Many DoubleML techniques support multiple treatments. For example, this is supported in EconML DML classes. Are there any plans to add support for multiple treatments in the SynapseML Causal ML library?

**Additional context**
This would help make the causal ML tools practical in the real world - most non-academic problems have multiple treatment variables.

Start by reading the SynapseML Causal ML library and checking how treatment variables are currently represented and limited to one treatment. Compare that behavior with the multiple-treatment support described for EconML and DoubleML; done would require an agreed design and documented support for multiple treatments at larger data sizes.

Written by the indexing model from the issue text.