Feature: add per-plan hardware performance counters to `EXPLAIN PERF`

[dqhl76]

Description:

Integrate Linux perf_event hardware counters into the pipeline executor to provide CPU microarchitecture-level insights for each plan node in EXPLAIN PERF output.

Motivation:

EXPLAIN PERF currently provides flamegraph-based CPU profiling (via sampling), which shows where time is spent but not why. Hardware performance counters reveal the root cause — whether a processor is bottlenecked by memory latency, branch mispredictions, or instruction throughput — enabling targeted
optimization without external profiling tools.

[dqhl76]

Restrictions

1. can only used on Linux environment (depends on perf_event)
2. perf_event_paranoid > 2 cannot work due to security policy