Enable users to utilize tokens abstraction in the converted tokens to prevent multiple LoRAs collisions.

**Is your feature request related to a problem? Please describe.**
By I guess mostly historical reasons the token is being used for the provided entity, which is fine, unless I load multiple LoRAs that use the same token. How do we distinguish which one entity will get called?

**Describe the solution you'd like.**
I propose we make an experimental flag that will utilize provided token abstractions instead of `s`, for example . In case of mulitple tokens that would be , , ...