Submit batch evaluation run on PR

[jjgriff93]

https://learn.microsoft.com/en-us/azure/ai-studio/how-to/flow-bulk-test-evaluation

What is the implication of dynamic guideline configurations being implemented in #19?

We might need to version guidelines (agents) in the db and have them as inputs to the flow so we can specify versions as variants when running eval. Or we have a separate non-dynamic flow node (similar to the concept of the streaming flag) that we call with dynamic=disabled that ignores fetching guidelines from Cosmos and uses local files (jinja templates) instead. This could also be used in Test and Prod environment to keep fixed prompts defined in files.