This PR gives the float_ function protected status with respect to operations from the pattern matcher and the transform and argument statements. For example,

CFunction f;
Symbol a;
Local F = 1.0;
id f?(?a) = g(?a);

does not match to the float_ function.

Furthermore, all statements that rely on the same compiler and execution routines -- DoArgument and execarg -- such as
argument, normalize, makeinteger, splitarg, factarg, and argtoextrasymbol, respect the same protected status of float_.

Note: at present, the protected status of float_ remains active even after the floating-point system has been closed using #endfloat. This behaviour may be relaxed in the future.