[Feat]Support Additional Load Testing Frameworks

[wangchen615]

Description

Expand fmperf's load testing capabilities by integrating additional load testing frameworks to provide more comprehensive benchmarking options.

Target Frameworks

• GuideLLM
• LMCache/LMBenchmark
• NVIDIA GenAIPerf
• Upstream vLLM benchmarking script

Implementation Requirements

1. Framework Integration

   • Define workload specifications for each framework
   • Support for different request patterns
   • Handle framework-specific configurations

2. Workload Generation

   • Support for various input/output token distributions
   • Handle different request types (chat, completion, etc.)
   • Support for custom workload patterns

3. Results Collection

   • Standardize metrics collection across frameworks
   • Support for framework-specific metrics
   • Handle different output formats

Expected Benefits

• More comprehensive benchmarking capabilities
• Support for different testing scenarios
• Better comparison between different load testing approaches

Technical Considerations

• Need to handle different API formats
• Consider different concurrency models
• Account for framework-specific limitations

Related Components

• fmperf/WorkloadSpecs.py
• Load testing configuration files
• Results processing utilities