fs/resctrl: Temp fix for truncated domain id

arm64 doesn't have ids for caches, it has to generate them from the
topology information. This means a cache-id, and thus domain-id
could be as much as 32bits. Increase the size of the bitfield to
allow this.

A better fix would be to get arm64 to compact its cache-ids into
the range 0..1..2 that user-space (probably) expects. This involves
sorting a list...

domain id --> unique id of the resource and on ARM64
this could be 32 bit.

This is a temp fix for,

estuary:/$ ./resctrl_tests -t cqm
TAP version 13
ok kernel supports resctrl filesystem
ok resctrl mountpoint "/sys/fs/resctrl" exists
ok mounting resctrl to "/sys/fs/resctrl"
cbm_mask_path /sys/fs/resctrl/info/L3/cbm_mask
cbm_mask 7fff
cache_path /sys/bus/cpu/devices/cpu1/cache/index3/size
cache_str 32768K
ok writing benchmark parameters to resctrl FS
get_resource_id: resource_id 524288
set_cqm_path: ctrlgrp c1 mongrp m1 sock_num 524288
get_llc_occu_resctrl: llc occupancy file /sys/fs/resctrl/c1/mon_groups/m1/mon_data/mon_L3_524288/llc_occupancy opened
Could not get llc occupancy: No such file or directory
not ok CQM: test
1..5
estuary:/$

Signed-off-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
Signed-off-by: James Morse <james.morse@arm.com>