Evaluate antonmedv/expr for use in filters and transforms

[punya]

#5680 introduces a DSL for filtering logs, and based on prior discussions we want to consider a similar approach for metric transforms/filters too. Before committing to this approach and this library, we'd like to do a deeper evaluation:

• overall performance impact of evaluating DSL at runtime
• comparison with other DSL approaches
• can we handle filter for resource && inst lib && signal without evaluating resource multiple times

Thanks @bogdandrutu for the detailed suggestion. Please add any other evaluation criteria that are on your mind.

[bogdandrutu]

I made a prototype in the past using other library open-telemetry/opentelemetry-collector#2408

[pmalek-sumo]

Is there anyone working on this? Have we done any of the research mentioned in the description?

[punya]

My apologies for not working on this issue. However, in light of @anuraaga's work on a generic transform processor (design doc, WIP implementation), I believe this evaluation is no longer necessary.