Add optional prompt template transformation in input transform

[ohltyler]

Description

For RAG use cases, users may want to configure the prompt templates directly in the ML processors, dynamically using data inputs as part of the prompt. For example, collecting a list of documents as context to pass to an LLM to summarize and return a human-readable response.

This PR adds a component in the input transform modal to view the original prompt template, and the prompt template with any injected input values (if applicable).

More details:

• adds prompt-related state vars and custom hooks when a prompt and/or input parameters are populated in InputTransformModal
• dynamically renders the new components if a valid prompt field is found in the processor's model_config field
• allows toggling between showing the prompt or not
• allows toggling between showing the original prompt, or the transformed version, with injected parameter values
• disables toggling and auto-selects options based on the transformed options being populated or not
• also: defaults all of the JSON fields to have wrap=true so long values (e.g., prompts) don't stretch out on one line
• also: minor var name updates to be consistent based on input/output contexts

Note this is just a first iteration at exposing the prompt template in the transform modal, and being able to see the finally-produced template that the model will receive. It will likely change, handle more edge case states, handle direct prompt manipulation, more details, etc. in the future.

Demo video, showing the prompt template being populated with parameter values:

screen-capture.22.webm

Issues resolved

Makes progress on #380

Check List

• Commits are signed per the DCO using --signoff

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.