Round-trip OTLP-OTAP-OTLP validation framework

[jmacd]

The goal is to test the OTAP receiver logic originally contributed in otel-arrow-rust using the pdata library to build OTLP objects for readable tests.

Part of #338
Part of #353

[jmacd]

Reviewers, I believe I've addressed all the feedback.

• Structured errors throughout (using snafu because of the precedent here, I will form an opinion about it vs thiserror)
• Convert to tokio::process::Command, remove use of threads and sync/async adapters
• Simplify and improve readability.

[jmacd]

I attempted to fix the otel-arrow-rust tests. The make otelarrowcol step needs to be run in the CI/CD workflow to build an otelarrowcol binary for the Rust tests (or we could use a docker image, etc). I think it will be best to merge this and diagnose the failures. Tests pass locally.

[jmacd]

I tried to address the formatting disagreement (between prost/tonic and rustfmt), which appears to be about editions. I tried to resolve that as well, but could not (see the new rustfmt.toml file).

FYI all clippy tests pass, edition is set to 2024, and tonic/prost are set to matching versions.