Fix FPU checks

[q009]

Compiling produces the following results:
http://pastebin.com/3R8KJG72
It appears it's related to the latest commit to this file, which adds a few preprocessor directives checking for an FPU. Oddly enough those directives used different syntax (they were prefixed with '#' instead of '.'), and I was suggested on IRC to change it so it complies with the rest.
This is the fix.

[adamgreen]

Another way to fix this problem would be to modify the GCC export scripts to use GCC to build assembly language sources instead of AS. If you specify the -x assembler-with-cpp flags to gcc, it will properly assemble this code after first running it through the C preprocessor. I see that this is used by workspace_tools/toolchains/gcc.py script so that is why the rtos still compiles when using the Python based build system.

Since the Python build system being used by Travis uses GCC for building the asm files, the build fails since the C preprocessor will replace the __FPU_PRESENT identifier in the code with either 0 or 1 before running the assembler. The values 0 and 1 aren't valid identifiers and therefore cause an error.

[0xc0170]

@q009 Did you use exported project to get the error you shared?

There's a pull request regarding gcc arm templates, which will be merged soon, once that gets merged, we could fix it.

@adamgreen Good catch with $(AS) $(CPU) -o $@ $< in the templates !

[q009]

Yes, I just pulled the project with a GCC ARM Embedded Makefile from mbed and it just gives those errors unless I apply a fix.

[0xc0170]

@q009 Thanks for the pull request. I'll close this, we should fix it as Adam pointed out - in the makefile templates. We should wait until the gcc template refactoring lands - #1159

[q009]

Awesome! Thanks for the assistance