Add support for LoRA in prompt tuning module (Investigate)

## Is your feature request related to a problem? Please describe.

Currently we only support vanilla prompt tuning and multi-task prompt tuning from PEFT library. LoRA has shown to give good performance on variety of datasets. This story is to investigate support for LoRA into the module.

Start by locating the prompt tuning module and its existing integrations with the PEFT library for vanilla and multi-task prompt tuning. Investigate what LoRA support would require and document a concrete scope, affected entry points, and validation approach; the issue does not name files or tests, so completion criteria remain to be established.

Written by the indexing model from the issue text.