Add bpf_task_state and bpf_pystack

[albertxu216]

bpf_task_state: a BPF-based tool for collecting and analyzing process state transitions, latency, memory usage, IO, and stack traces in Linux.
Features:

• Collects per-process and system-wide state statistics (running, sleeping, etc.)
• Supports filtering by PID and process name (comm), with wildcard and multi-process support
• Tracks memory usage, read/write bytes, and page delay metrics (FreePages, Thrashing, SwapIn)
• Outputs kernel and user stack traces for each process, with symbol resolution
• Supports interval and event-driven sampling, latency threshold filtering, and per-instance statistics
• Provides flame graph data for performance analysis

Usage examples:

• Show all process state statistics every second: perf-prof bpf:task-state -i 1000

• Filter by PID: perf-prof bpf:task-state -p 1234 -i 1000

• Filter by process name (supports wildcards): perf-prof bpf:task-state --filter 'java,python*' -i 1000

• Only show events with latency greater than 20ms and print stack traces: perf-prof bpf:task-state --filter 'java,python*' -SD --than 20ms -g

• Output per-instance statistics with latency threshold: perf-prof bpf:task-state --perins --than 1ms -i 1000

bpf-pystack: Uses eBPF technology to maintain the call stacks of Python threads；

• Outputs them python-stacks with kernel and user call stacks at appropriate times;
• Compared to pystack, bpf-pystack supports stack maintenance for newly created processes.