scx_p2dq Add scaled vtime to accounting

[hodgesds]

When weighted tasks are run use scaled vtime to appropriately account the vtime on the LLC/DSQ of the task.

@JakeHillion and I noticed some issues with vtime handling going through the code. There's still probably more to cleanup, but this is an easy fix.

Performance remains good with stress-ng tests:

sudo ./target/release/scx_p2dq --stats 1 -s 1000 -x 1:

$ stress-ng -c 60 -t 15 -M
stress-ng: info:  [394158] setting to a 15 secs run per stressor
stress-ng: info:  [394158] dispatching hogs: 60 cpu
stress-ng: metrc: [394158] stressor       bogo ops real time  usr time  sys time   bogo ops/s     bogo ops/s CPU used per       RSS Max
stress-ng: metrc: [394158]                           (secs)    (secs)    (secs)   (real time) (usr+sys time) instance (%)          (KB)
stress-ng: metrc: [394158] cpu              872017     15.00    893.49      0.10     58127.58         975.86        99.28          5248
stress-ng: info:  [394158] skipped: 0
stress-ng: info:  [394158] passed: 60: cpu (60)
stress-ng: info:  [394158] failed: 0
stress-ng: info:  [394158] metrics untrustworthy: 0
stress-ng: info:  [394158] successful run completed in 15.02 secs

eevdf:

$ stress-ng -c 60 -t 15 -M
stress-ng: info:  [390654] setting to a 15 secs run per stressor
stress-ng: info:  [390654] dispatching hogs: 60 cpu
stress-ng: metrc: [390654] stressor       bogo ops real time  usr time  sys time   bogo ops/s     bogo ops/s CPU used per       RSS Max
stress-ng: metrc: [390654]                           (secs)    (secs)    (secs)   (real time) (usr+sys time) instance (%)          (KB)
stress-ng: metrc: [390654] cpu              871792     15.00    894.13      0.15     58110.48         974.86        99.35          5268
stress-ng: info:  [390654] skipped: 0
stress-ng: info:  [390654] passed: 60: cpu (60)
stress-ng: info:  [390654] failed: 0
stress-ng: info:  [390654] metrics untrustworthy: 0
stress-ng: info:  [390654] successful run completed in 15.01 secs

[etsal]

Makes sense and is consistent with other schedulers.