Preparations for ARMv8.2

[marenz2569]

DoD

• Split command line parsing of --cpu-bind string into its own class. allow setting thread affinity based on this result. This is orthogonal to Replace CPU_SET calls with thread affinity from hwloc #74.
  • Done with Refactor thread affinity #96
• Issue Remove cross platform compatibility code for inline assembler #86
• Refactor function selection/function summary to support multiple architecture (ISA)
  • Done with Refactor Function Selection #98
• Split config parsing and code based on parsed config settings
• find a substitute for family/model for Arm
• Find a way to extract clockrate/timestamp on ARMv8 (timestamp is important for ipc-estimate metric and performance report)
• Refactor default/neon payload from prototype PR: master...aarch64_support
• Abstract memory fence between X86 and ARMv8, also select correct memory fence for ARMv8

Open Questions

• Should the function IDs start at 1 for each ISA (x86_64, armv8 etc.) or should they be in sequence (available function report will show also x86_64 functions on ARM?)