pp.h: Use STATIC_ASSERT_EXPR on SETs

[khwilliamson]

This makes sure that the argument whose pointer is being passed has the correct size to be settable by the called element.

• This set of changes does not require a perldelta entry.

[tonycoz]

When wouldn't this be caught by the compiler?

This could only be a problem when called from non-XS code (ie. no dSP explicit or implicit), since *SP is a SV*, s can only be a pointer (always the same size as SV* on anything we use) or an integer constant that evaluates to zero, which is valid but rare (same as NULL).

Non-SV pointers (which this change doesn't catch) give you:

<source>:11:5: warning: incompatible pointer types assigning to 'SV *' (aka 'struct sv *') from 'int *' [-Wincompatible-pointer-types]
   11 |     SETs(&x);
      |     ^    ~~
/opt/compiler-explorer/perl-5.42.2/lib/5.42.2/x86_64-linux-thread-multi/CORE/pp.h:583:23: note: expanded from macro 'SETs'
  583 | #define SETs(s)         (*sp = s)
      |                              ^ ~
1

Non pointers/non-integer:

<source>:11:5: error: assigning to 'SV *' (aka 'struct sv *') from incompatible type 'double'
   11 |     SETs(0.0);
      |     ^    ~~~
/opt/compiler-explorer/perl-5.42.2/lib/5.42.2/x86_64-linux-thread-multi/CORE/pp.h:583:23: note: expanded from macro 'SETs'
  583 | #define SETs(s)         (*sp = s)
      |                              ^ ~

clang complains about this valid but strange value (same as SETs(0)):

<source>:12:10: warning: expression which evaluates to zero treated as a null pointer constant of type 'SV *' (aka 'struct sv *') [-Wnon-literal-null-conversion]
   12 |     SETs('\0');
      |          ^~~~
/opt/compiler-explorer/perl-5.42.2/lib/5.42.2/x86_64-linux-thread-multi/CORE/pp.h:583:25: note: expanded from macro 'SETs'
  583 | #define SETs(s)         (*sp = s)
      |                                ^

(clang++/g++ report an error, since '\0' is a char in C++ and an int in C)

Tested here.