feat: Add support for BYO tracing_subscriber

[justinabrahms]

Allows exfiltration of instrumentation from noosphere into handy places like otel syncs like:

    let tracer = opentelemetry_otlp::new_pipeline()
        .tracing()
        .with_trace_config(
            trace::config().with_resource(Resource::new(vec![KeyValue::new(
                "service.name",
                "my service",
            )])),
        )
        .with_exporter(opentelemetry_otlp::new_exporter().tonic())
        .install_batch(opentelemetry_sdk::runtime::Tokio)?;

    let otel = tracing_opentelemetry::layer().with_tracer(tracer);

    initialize_tracing_with_layer(None, "multi-tenant gateway", otel);

[cdata]

This looks like a good step in the right direction, although I still think we want a pattern where Noosphere-specific tracing behavior is encapsulated as its own Layer, and configuration of the Subscriber happens outside of this code.

[github-actions]

Test flake analysis

status	platform	features	toolchain
🟢	macos-13	test-kubo,headers,rocksdb	stable
🟢	windows-latest	test-kubo,headers	stable
🟡	macos-13	test-kubo,headers	stable
🟢	ubuntu-latest	test-kubo,headers	stable
🟢	ubuntu-latest	test-kubo,headers,rocksdb	stable
🟢	ubuntu-latest	test-kubo,headers	nightly

Flake summary for macos-13, test-kubo,headers, stable

     Summary [ 127.663s] 176 tests run: 176 passed (1 flaky), 1 skipped
   FLAKY 2/5 [  52.103s] noosphere::distributed_stress latency::clients_can_sync_when_there_is_a_lot_of_content

[jsantell]

With the wrangling of the multiple layer types, this is a nice & simple way of extending functionality. LGTM with lint passing!