various eap-aka-prime fixes

[edipascale]

• remove spurious length check when fetching attributes, such as in getRand(), getMac() etc.
• fix the AT_RES length (it's in bits, not bytes)
• fix reordering of attributes and MAC computation issues
• fix SUPI input to the master key (remove the supi type prefix)
• use kAusf computation for eap-aka' rather than the 5g-aka one

Inspired in large part by the issues and solutions reported in #592