feat: add multi-spec execution with result aggregation and timeseries replay

[JasonXuDeveloper]

Summary

Add support for executing multiple specs sequentially within a single kperf runner invocation, enabling time-series replay scenarios where workload patterns vary over time.

Changes

Core Multi-Spec Functionality

• Add MultiSpecRunnerMetricReport type for multi-spec results in api/types/metric.go
• Implement executeSpecs() for sequential spec execution
• Implement aggregateResults() to combine per-spec results into aggregated metrics
• Add printMultiSpecResults() for multi-spec output format
• Add hasCliOverrides() to detect and prevent CLI overrides with multi-spec configs
• Update buildRunnerGroupSummary() in runner/utils.go to handle both single and multi-spec report formats

Example and Demo

• Add timeseries replay example config with 3 phases:
  • Baseline: 10 QPS for 30s
  • Spike: 100 QPS for 20s
  • Recovery: 25 QPS for 30s
• Add benchTimeseriesReplayCase command to demonstrate time-series replay

Technical Details

• Multi-spec configs with CLI flag overrides are rejected with clear error message
• Single-spec configs continue to work without changes
• All specs share the same HTTP client pool (configured from first spec)
• Results are aggregated with per-spec breakdown and overall summary

Use Cases

This enables realistic workload patterns useful for:

• Stress testing with gradual load increases
• Capacity planning with varying traffic patterns
• Validating system behavior during load spikes and recovery
• Time-series replay of production traffic patterns

Testing

• ✅ Built and tested locally
• ✅ All existing single-spec configs work unchanged
• ✅ Multi-spec timeseries replay example validated
• ✅ Proper aggregation of results across multiple specs

Related

This PR builds on #226 which converted the Spec field to Specs array format.