Phase 2: Implement Execution Environment Abstraction (v0.3.0)

This commit implements Phase 2 of the CHORUS Task Execution Engine development plan,
providing a comprehensive execution environment abstraction layer with Docker
container sandboxing support.

## New Features

### Core Sandbox Interface
- Comprehensive ExecutionSandbox interface with isolated task execution
- Support for command execution, file I/O, environment management
- Resource usage monitoring and sandbox lifecycle management
- Standardized error handling with SandboxError types and categories

### Docker Container Sandbox Implementation
- Full Docker API integration with secure container creation
- Transparent repository mounting with configurable read/write access
- Advanced security policies with capability dropping and privilege controls
- Comprehensive resource limits (CPU, memory, disk, processes, file handles)
- Support for tmpfs mounts, masked paths, and read-only bind mounts
- Container lifecycle management with proper cleanup and health monitoring

### Security & Resource Management
- Configurable security policies with SELinux, AppArmor, and Seccomp support
- Fine-grained capability management with secure defaults
- Network isolation options with configurable DNS and proxy settings
- Resource monitoring with real-time CPU, memory, and network usage tracking
- Comprehensive ulimits configuration for process and file handle limits

### Repository Integration
- Seamless repository mounting from local paths to container workspaces
- Git configuration support with user credentials and global settings
- File inclusion/exclusion patterns for selective repository access
- Configurable permissions and ownership for mounted repositories

### Testing Infrastructure
- Comprehensive test suite with 60+ test cases covering all functionality
- Docker integration tests with Alpine Linux containers (skipped in short mode)
- Mock sandbox implementation for unit testing without Docker dependencies
- Security policy validation tests with read-only filesystem enforcement
- Resource usage monitoring and cleanup verification tests

## Technical Details

### Dependencies Added
- github.com/docker/docker v28.4.0+incompatible - Docker API client
- github.com/docker/go-connections v0.6.0 - Docker connection utilities
- github.com/docker/go-units v0.5.0 - Docker units and formatting
- Associated Docker API dependencies for complete container management

### Architecture
- Interface-driven design enabling multiple sandbox implementations
- Comprehensive configuration structures for all sandbox aspects
- Resource usage tracking with detailed metrics collection
- Error handling with retryable error classification
- Proper cleanup and resource management throughout sandbox lifecycle

### Compatibility
- Maintains backward compatibility with existing CHORUS architecture
- Designed for future integration with Phase 3 Core Task Execution Engine
- Extensible design supporting additional sandbox implementations (VM, process)

This Phase 2 implementation provides the foundation for secure, isolated task
execution that will be integrated with the AI model providers from Phase 1
in the upcoming Phase 3 development.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

vendor/go.opentelemetry.io/otel/internal/global/trace.go

@@ -1,16 +1,5 @@
 // Copyright The OpenTelemetry Authors
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
+// SPDX-License-Identifier: Apache-2.0
 
 package global // import "go.opentelemetry.io/otel/internal/global"
 
@@ -36,9 +25,12 @@ import (
 	"sync"
 	"sync/atomic"
 
+	"go.opentelemetry.io/auto/sdk"
+
 	"go.opentelemetry.io/otel/attribute"
 	"go.opentelemetry.io/otel/codes"
 	"go.opentelemetry.io/otel/trace"
+	"go.opentelemetry.io/otel/trace/embedded"
 )
 
 // tracerProvider is a placeholder for a configured SDK TracerProvider.
@@ -46,6 +38,8 @@ import (
 // All TracerProvider functionality is forwarded to a delegate once
 // configured.
 type tracerProvider struct {
+	embedded.TracerProvider
+
 	mtx      sync.Mutex
 	tracers  map[il]*tracer
 	delegate trace.TracerProvider
@@ -94,6 +88,8 @@ func (p *tracerProvider) Tracer(name string, opts ...trace.TracerOption) trace.T
 	key := il{
 		name:    name,
 		version: c.InstrumentationVersion(),
+		schema:  c.SchemaURL(),
+		attrs:   c.InstrumentationAttributes(),
 	}
 
 	if p.tracers == nil {
@@ -112,6 +108,8 @@ func (p *tracerProvider) Tracer(name string, opts ...trace.TracerOption) trace.T
 type il struct {
 	name    string
 	version string
+	schema  string
+	attrs   attribute.Set
 }
 
 // tracer is a placeholder for a trace.Tracer.
@@ -119,6 +117,8 @@ type il struct {
 // All Tracer functionality is forwarded to a delegate once configured.
 // Otherwise, all functionality is forwarded to a NoopTracer.
 type tracer struct {
+	embedded.Tracer
+
 	name     string
 	opts     []trace.TracerOption
 	provider *tracerProvider
@@ -147,6 +147,41 @@ func (t *tracer) Start(ctx context.Context, name string, opts ...trace.SpanStart
 		return delegate.(trace.Tracer).Start(ctx, name, opts...)
 	}
 
+	return t.newSpan(ctx, autoInstEnabled, name, opts)
+}
+
+// autoInstEnabled determines if the auto-instrumentation SDK span is returned
+// from the tracer when not backed by a delegate and auto-instrumentation has
+// attached to this process.
+//
+// The auto-instrumentation is expected to overwrite this value to true when it
+// attaches. By default, this will point to false and mean a tracer will return
+// a nonRecordingSpan by default.
+var autoInstEnabled = new(bool)
+
+// newSpan is called by tracer.Start so auto-instrumentation can attach an eBPF
+// uprobe to this code.
+//
+// "noinline" pragma prevents the method from ever being inlined.
+//
+//go:noinline
+func (t *tracer) newSpan(
+	ctx context.Context,
+	autoSpan *bool,
+	name string,
+	opts []trace.SpanStartOption,
+) (context.Context, trace.Span) {
+	// autoInstEnabled is passed to newSpan via the autoSpan parameter. This is
+	// so the auto-instrumentation can define a uprobe for (*t).newSpan and be
+	// provided with the address of the bool autoInstEnabled points to. It
+	// needs to be a parameter so that pointer can be reliably determined, it
+	// should not be read from the global.
+
+	if *autoSpan {
+		tracer := sdk.TracerProvider().Tracer(t.name, t.opts...)
+		return tracer.Start(ctx, name, opts...)
+	}
+
 	s := nonRecordingSpan{sc: trace.SpanContextFromContext(ctx), tracer: t}
 	ctx = trace.ContextWithSpan(ctx, s)
 	return ctx, s
@@ -156,6 +191,8 @@ func (t *tracer) Start(ctx context.Context, name string, opts ...trace.SpanStart
 // SpanContext. It performs no operations other than to return the wrapped
 // SpanContext.
 type nonRecordingSpan struct {
+	embedded.Span
+
 	sc     trace.SpanContext
 	tracer *tracer
 }
@@ -186,6 +223,9 @@ func (nonRecordingSpan) RecordError(error, ...trace.EventOption) {}
 // AddEvent does nothing.
 func (nonRecordingSpan) AddEvent(string, ...trace.EventOption) {}
 
+// AddLink does nothing.
+func (nonRecordingSpan) AddLink(trace.Link) {}
+
 // SetName does nothing.
 func (nonRecordingSpan) SetName(string) {}