The macro is (putatively) more efficient, as it doesn’t involve a function call. It can be optimised more easily, as it just involves a pointer offset lookup.

The function call allows linking against the same library even if the program was compiled without the macro definition – if it was compiled with a different header, or just with a rogue declaration inside the source file. Should, for example, you have a compiler which has someone’s “improved” version of ctype.h that didn’t have the macro, the function would still exist at runtime for use.

If we look at the standard:

c99

7.1.4 Use of library functions

Any function
declared in a header may be additionally implemented as a function-like macro defined in the header, so if a library function is declared explicitly when its header is included, one
of the techniques shown below can be used to ensure the declaration is not affected by
such a macro. Any macro definition of a function can be suppressed locally by enclosing
the name of the function in parentheses, because the name is then not followed by the left
parenthesis that indicates expansion of a macro function name. For the same syntactic
reason, it is permitted to take the address of a library function even if it is also defined as a macro.

That means that if you write:

int b = (isdigit)(c);

or

int (*f)(int) = &isdigit;
int b = f(c);

then you are invoking the actual function, not the macro. You can also legally write:

#undef isdigit
int b = isdigit(c);

or (in a source file not having #include <ctype.h> directly or transitively):

extern int isdigit(int);
int b = isdigit(c);