Add opt-in OTLP/HTTP agent tracing

[itsperini]

Hi! I’ve been trying Amplio locally and wanted to inspect model/tool latency, token usage, and failures in an existing observability setup.

This adds optional OTLP/HTTP tracing using standard OpenTelemetry configuration. It’s disabled by default, with asynchronous batching and no backend-specific integration. Amplio’s spans exclude prompts, responses, and tool contents.

• Step, model, and tool spans with run/session correlation
• Streaming usage, error categories, and bounded shutdown flushing
• Configuration docs and dependency license notices
• Full race-test suite, lint, build, and a real model run exporting to a local OTLP receiver
• Discuss span boundaries and attributes with maintainers; address feedback before marking ready

Most added lines are license notices and tests. Opening this as a draft to get early feedback, happy to adjust the scope or discuss other ideas!

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.