Implemented float.h

and improved math.h

libc/math/const.c

@@ -1,4 +1,5 @@
 #include <math.h>
+#include <string.h>
 
 double nan(const char* tagp) {
     (void)tagp;

libc/math/htrig.c

@@ -11,6 +11,6 @@ double cosh(double x) {
 double tanh(double x) {
     double ex = exp(x);
     double emx = exp(-x);
-    return (ex - exp) / (ex + exp);
+    return (ex - emx) / (ex + emx);
 }