Set correct float abi (float abi double) for hard-float build, so
__riscv_float_abi_double macro will be defined by compiler.

The options are:
o lp64 riscv_float_abi_soft
o lp64f riscv_float_abi_single
o lp64d __riscv_float_abi_double

Sponsored by: DARPA, AFRL