Add a function to implement the FP comparison macros in <math.h>.

These macros differ from the normal comparison operators in that they will not signal invalid due to the operands being unordered. However, this implementation will still signal invalid for SNaNs. That is clearly OK according to the wording in draft C23. C17 and earlier do not mention that possibility, but they do not really specify the behavior of SNaNs in general.

math2.asm

@@ -153,3 +153,44 @@ __signbit start
          
          creturn 2:val
          end
+
+****************************************************************
+*
+*  int __fpcompare(long double x, long double y, short mask);
+*
+*  Compare two floating-point values, not signaling invalid
+*  if they are unordered.
+*
+*  Inputs:
+*        x,y - values to compare
+*        mask - mask of bits as returned in X register from FCMP
+*
+*  Outputs:
+*        1 if x and y have one of the relations specified by mask
+*        0 otherwise
+*
+****************************************************************
+*
+__fpcompare start
+
+         csubroutine (10:x,10:y,2:mask),0
+         
+         tdc
+         clc
+         adc   #x
+         pea   0
+         pha
+         tdc
+         clc
+         adc   #y
+         pea   0
+         pha
+         FCMPX
+         txa
+         and   mask
+         beq   lb1
+         lda   #1
+lb1      sta   mask
+         
+         creturn 2:mask
+         end

math2.macros

@@ -121,3 +121,9 @@
  LDX #$090A
  JSL $E10000
  MEND
+ MACRO
+&LAB FCMPX
+&LAB PEA $08
+ LDX #$090A
+ JSL $E10000
+ MEND