Add support for FPv4-SP to the clang driver

[Tim Northover]

Hi Oliver,

> This patch adds support for the FPv4-SP floating point unit (an optional
> feature of Cortex-M4) to the clang driver. It will now accept two new
> options for –mfpu: “fpv4-sp-d16” and “fp4-sp-d16”. The first is the same
> spelling as gcc’s option, and the second is for consistency with the other
> clang options which allow dropping the “v”. It is correct that this option
> does not have a leading “v”, this is the case in both ARM’s documentation
> and gcc’s implementation of the option.

Just the points I was about to ask about! Looks good to me.

Tim.