Bring up x86 SMP and per-core runtime telemetry
* Add x86 SMP bring-up end-to-end: reserve a low trampoline page in BIOS/UEFI, boot APs via LAPIC INIT/SIPI trampolines, initialize per-core APIC timer/IDT/GDT/FPU state, and enable TLB shootdown IPIs.
* Rework scheduler/core plumbing for multicore startup with BSP/AP handoff, per-core usage counters, runnable affinity handling, poll wait wakeups, and stronger guards against contextless or non-ready runqueue entries.
* Expand kernel process and CPU observability by exporting online core count plus per-core tick counters in /dev/cpu, adding core_id and vm_kib to /proc/<pid>/stat, tracking user VM KiB through mmap/munmap/exec, and locking procfs directory walks safely.
* Refresh userland diagnostics/runtime defaults: teach top/fetch the new CPU/memory/core fields (including optional per-core and bar views), pass QEMU SMP count via QEMU_SMP, and adjust Doom build optimization flag handling.