RACF-KDFAES tunable costs

Apparently, RACF-KDFES has at least 2 tunable costs, mfact and rfact, which are currently not reported.
And, apparently, for benchmarking we should ideally have test vectors with lower cost values.

$ date; time ./john --test=0 --format=RACF-KDFAES; date; time ./john --test --format=RACF-KDFAES; date; time ./john --test=-1 --format=RACF-KDFAES; date
Mo 14. Jun 01:20:50 CEST 2021
Will run 8 OpenMP threads
Testing: RACF-KDFAES [KDFAES (DES + HMAC-SHA256/64 + AES-256)]... (8xOMP) PASS

real	0m32.974s
user	1m40.352s
sys	0m0.048s
Mo 14. Jun 01:21:23 CEST 2021
Will run 8 OpenMP threads
Benchmarking: RACF-KDFAES [KDFAES (DES + HMAC-SHA256/64 + AES-256)]... (8xOMP) DONE
Warning: "Many salts" test limited: 2/256
Many salts:	0.6 c/s real, 0.0 c/s virtual
Only one salt:	0.3 c/s real, 0.0 c/s virtual


real	1m23.371s
user	8m11.868s
sys	0m0.059s
Mo 14. Jun 01:22:47 CEST 2021
Will run 8 OpenMP threads
Benchmarking: RACF-KDFAES [KDFAES (DES + HMAC-SHA256/64 + AES-256)]... (8xOMP) DONE
Many salts:	0.6 c/s real, 0.0 c/s virtual
Only one salt:	0.3 c/s real, 0.0 c/s virtual


real	51m55.669s
user	410m38.680s
sys	0m0.283s
Mo 14. Jun 02:14:42 CEST 2021

This is just crazy.
I knew --test=-1 would be pointless, given that at this speed (and john reporting just on decimal place for c/s), I would hardly see any difference.
But I wanted to see just how long it would take.

Start from the --format=RACF-KDFAES entry point and inspect how its tunable costs and benchmark test vectors are represented. Done means the mfact and rfact costs are reported and lower-cost vectors make benchmarking practical without changing the format's behavior.