Reduce panic-driven internal control flow

[AlexandreYang]

Summary

Converts internal panics in hot paths to explicit fatal errors, reserving panics for truly unrecoverable conditions (public API misuse, user handler crashes).

Changes

• interp/vars.go: Added internalErrorf helper that records assertion failures via exit.fatal instead of panicking. Converted panics in lookupVar, setVarWithIndex, and assignVal to use it.
• interp/api.go: Converted Reset() panic for zero-value Runner to exit.fatal + early return. Added explicit fatal-error check in Run() after Reset() so the error surfaces without relying on the recover wrapper.
• interp/handler.go: Kept HandlerCtx panic (public API, genuine programmer error) but clarified the doc comment.
• interp/allowed_paths_internal_test.go: Added TestRunZeroValueRunnerReturnsError to verify the new explicit error path.

The recover() in Run() is retained as a safety net for panics from user-provided handlers (validated by existing TestRunRecoversPanic).

Testing

• All existing tests pass (go test ./...)
• New test verifies zero-value Runner returns an explicit error instead of panicking

---

PR by Bits
View session in Datadog

Comment @DataDog to request changes

[datadog-datadog-prod-us1-2]

View session in Datadog

Bits Dev status: ✅ Done

CI Auto-fix: Disabled | Enable

Comment @DataDog to request changes