Name ACPI hardware IDs instead of magic _HID strings

Turn acpi-ids.zig's flat name table into a HardwareId enum modeled on
ps2-library's Port: one entry() switch holds the registry (variant ->
_HID string + human-readable name), with hid(), description(), and
fromHid() methods. The free description(hid) lookup the kernel's
device-tree dump uses survives, implemented over the enum, and a new
test round-trips every variant through fromHid.

Callers now name the device instead of quoting its id:

- ps2-library's DeviceType.hid() and ps2-bus's descriptor lookups use
  HardwareId.ps2_keyboard / .ps2_mouse.
- device-manager's driverFor parses the HID once with fromHid and
  switches on named values.
- acpi.zig's isPciRootNode carried the same ids twice, as strings and
  as packed-EISA integers (0x030AD041/0x080AD041); both branches now
  decode to the string form and answer through one isPciRootHid helper
  using .pci_bus / .pci_express_root_bridge.
- build.zig threads the acpi-ids module (previously kernel-only) into
  every user binary, like xkeyboard-config.

system/services/device-manager/device-manager.zig

@@ -14,6 +14,7 @@
 //! that redundancy so the manager is the sole owner of driver spawning.)
 const std = @import("std");
 const runtime = @import("runtime");
+const acpi_ids = @import("acpi-ids");
 const device = runtime.device;
 const system = runtime.system;
 
@@ -34,9 +35,11 @@ fn driverFor(d: device.DeviceDescriptor) ?[]const u8 {
     if (d.class == @intFromEnum(device.DeviceClass.timer)) return "hpet";
     // detect device via hid
     const hid = d.hid[0..@intCast(d.hid_len)];
-    if (std.mem.eql(u8, hid, "PNP0303") or std.mem.eql(u8, hid, "PNP0F13")) return "ps2-bus";
-
-    return null;
+    const id = acpi_ids.HardwareId.fromHid(hid) orelse return null;
+    return switch (id) {
+        .ps2_keyboard, .ps2_mouse => "ps2-bus",
+        else => null,
+    };
 }