libdvbcsa: use enable-neon for Aarch64

[kszaq]

I have performed tests on Aarch64, WeTek Hub:

• --enable-uint64 gives ~200 Mbits/s
• --enable-neon gives ~215 Mbits/s

Ping @glenvt18

[glenvt18]

@kszaq Looks good. Thanks for testing. The difference in performance between --enable-uint64 and --enable-neon on cortex-a53 is small because the neon unit is only 64-bit (the same as on cortex-a7). On more powerful AArch64 cores with a 128-bit neon unit --enable-neon should be considerably faster.

BTW --enable-ssse3 is generally up to 15% faster on all x86_64 platforms but PROJECT_CFLAGS for Generic target doesn't allow it for the moment.