Apply correct, optimized mfpu compiler flag for ARMv7-A Cortex CPUs

For those ARMv7-A Cortex CPUs that can handle VFPv4 floating point,
We can set "-mfpu=neon-vfpv4" instead of generic "-mfpu=neon"
to gain extra performance improvements.

References :
 - GCC		: https://gcc.gnu.org/onlinedocs/gcc/ARM-Options.html
 - Cortex A15	: http://www.arm.com/products/processors/cortex-a/cortex-a15.php
 - Cortex A9	: http://www.arm.com/products/processors/cortex-a/cortex-a9.php
 - Cortex A8	: http://www.arm.com/products/processors/cortex-a/cortex-a8.php
 - Cortex A7	: http://www.arm.com/products/processors/cortex-a/cortex-a7.php
 - Cortex A5	: http://www.arm.com/products/processors/cortex-a/cortex-a5.php



Currently prebuilt clang doesn't support neon-vfpv4 flag so fall back to neon
Also added support for cortex-a5 arch

Change-Id: I91893789ed8edabf3767e1782e494b81158332bb
Signed-off-by: Park Ju Hyung <qkrwngud825@gmail.com>
Signed-off-by: Maxime Poulain <maxime.poulain@student.uclouvain.be>