Return (a superset of) EvaluationError from TPE evaluation

[luxas]

Category

User level API features/changes

Describe the feature you'd like to request

When evaluating a policy in TPE mode, normal evaluation errors might happen. For example, an arithmetic overflow can happen at any point. Right now, the context on what part of the evaluation failed is not exposed as part of Residual::Error. In order to get the same diagnosis possibilities as for normal evaluation, it'd be nice to get the same error causes exposed.

I was discussing with @shaobo-he-aws on Slack about this as well.

Describe alternatives you've considered

Additional context

No response

Is this something that you'd be interested in working on?

• 👋 I may be able to implement this feature request
• ⚠️ This feature might incur a breaking change

[oflatt]

Once this is done, we should also change batched evaluation to return a full concrete Response including Diagnostics instead of just a Decision.

[luxas]

I started working on this in #2162