feat(sampling): datadog sampler

[ZStriker19]

What does this PR do?

This PR adds the DatadogSampler, which has the following cap

Add DatadogSampler for intelligent trace sampling in dd-trace-rs

Summary

This PR adds a comprehensive sampling implementation for the Datadog tracer in Rust. The DatadogSampler provides probabilistic and rule-based sampling capabilities to selectively record traces based on configurable criteria.

Features

• Rule-based sampling: Define sampling rules based on service name, operation name, resource name, and custom tags
• Probabilistic sampling: Apply different sampling rates to different services and operations
• Service rate sampling: Support for dynamic sampling rates based on service/environment combinations
• Rate limiting: Prevent excessive trace generation with configurable rate limits
• Priority sampling: Full support for Datadog's sampling priority mechanism
• Glob pattern matching: Use wildcards in rule definitions for flexible matching
• Proper context propagation: Consistent sampling decisions throughout trace hierarchies
• JSON configuration: Configure the sampler via structured JSON input
• Initialize the sampler with configuration and pass it to the tracer
• Checks span attributes according to Datadog semantic conversions (will refactor these constants out in next PR)

Implementation details

The implementation consists of several cooperating components:

• DatadogSampler: The main sampler implementing OpenTelemetry's ShouldSample trait
• SamplingRule: Configurable rules for matching spans and applying sampling decisions
• RateSampler: Deterministic sampling based on trace ID hash values
• RateLimiter: Thread-safe limiting of sampling throughput
• GlobMatcher: Fast pattern matching with wildcard support

The sampler is fully integrated with the Datadog OpenTelemetry tracer, which can be initialized with:

let sampler = DatadogSampler::new(None, None);
let tracer_provider = tracer_provider_builder
    .with_sampler(sampler)
    .build();

Testing

The implementation includes a comprehensive test suite covering all aspects of the sampler's behavior, including rule matching, sampling decisions, tag generation, and configuration loading.

[bantonsson]

I'm only halfway through this, but here are some comments.

[bantonsson]

Looks good AFAICT. Only nitpicks and performance questions.

[paullegranddc]

Should this be an enumeration if the values this can take is bounded?
Also where does this come from, can customer provide this?

[iunanua]

Handling agent response is missing and Sampler configuration can be modified by agent (see TraceExporterWorker::handle_agent_reponse)

Is it planned for another PR?

[paullegranddc]

Is it planned for another PR?

Yes, the changes on this one are already big enough 😄