The argument "0" is not automatically converted to the right type unless there is a prototype in scope. It isn't as important in this case because it is highly likely that the appropriate prototype has been #included, but it is a bigger deal if we're dealing with arguments for a variadic function. Anyway, it's good to be reminded what the declared types are.
Why would the utilities not handle unicode searching? Unicode characters match properly, the null terminator works the same, and non-ANSI codes are just one or more random 8-bit values which can be compared, copied, etc.
In fact I proposed strdup on a few occasions, but it wasn't adopted. It seems that they didn't like for standard library functions to use malloc.
POSIX.1 specifies strdup.
If you changed UTF-16 to UTF-32 or UCS-4 I'd support it. I think there are already implementations that use the replacement character for all "impossible" codes.
I don't think most users of C want things changing underfoot.
Keeping track of all the version combinations is infeasible, especially when you consider that an app and its library packages are likely to have been developed and tested for a variety of environments. To the extent that existing correct code has to be scanned and revised when a new compiler release comes out, one of the primary goals of standardization has failed.
Code can be non-portable for various reasons, not all of them bad. I just grepped a recent release of DWB and found about 100 uses of isdigit, most of which were not input from random text but rather were used internally, such as "register" names (limited to a specified range). Other packages are likely to have similar usage patterns. I really don't want to have to edit that code just for aesthetics.
Many antique computers are simulated by SIMH. If you have the corresponding software, you can operate on your desktop a simulated computer's software development system. For example, DEC VAX (VMS or Unix) has a relatively simple and sane assembly language.
(1) There are several implementations; most are based on Knuth's "boundary tag" algorithms. As to "heap", a stack has one accessible end, a heap is essentially random-accessible. Nothing to do with the heap data structure.
(2) Stack overflow can occur even early within a program. I've campaigned for a requirement that such overflows be caught and integrated into a standard exception handler, to no avail. (3) Why not code your own, so there won't be arguments about it. (4) There are lots of tools for program development, but it's not standardized by WG14. (5) Use wider integer types. (6) Use wider floating representations. (7) Standard C doesn't specify such a facility, but it has occasionally be suggested. (8) There were a lot of books, e.g. on structured system analysis, during the 1970s trying to apply lessons learned. C isn't special in that regard, as many of the big problems don't involve syntax. (9) C++ is now a big language and it takes a lot of work to master its internals.
Note that the ABIs cover endianness as well as value range and/or object widths. In general, one needs to have explicit marshaling and unmarshaling functions to map from network octet array and C internal data representation.
Failure to get this right is (or used to be) a common bug for code developed and tested on too few architectures.
If you're using parentheses, as has been recommended for decades, there is no problem. Otherwise, it is likely that such a change would adversely impact previously working code. There just isn't a pressing need to change it.
Let's assume the types have been corrected.
malloc((size_t)0) behavior is defined by the implementation; there are two choices: (a) always returns a null pointer; or (b) acts like malloc((size_t)1) which can allocate or fail, and if it allocates then the program shall not try to reference anything through the returned non-null pointer.
Now, memset itself is required (among other things) to be given as its first argument a valid pointer to a byte array. In particular, it shall not be a null pointer. Tracking through the conformance requirements, if the malloc call returns a null pointer then the behavior is undefined. Thus, you should not program like this.
"Implementation-defined" is a nuisance, because then you need to add code for all the variations, which also requires a set of standard macros, etc. It is easier and less trouble-prone to just avoid using the currently undefined behavior.
You can easily create your own stdc.h include file. Something similar was done on Plan 9.
Note that by including the content of all the headers, you're increasing the chance for collisions with application identifiers. You might consider that more of a benefit than a drawback.
There are a lot of arithmetic conditions for which C could generate special code. There are div_t-related functions for the other direction. I for one would like a good way to obtain, using some Standard C coding pattern, fast "carry" for multiple-precision integer arithmetic.
Several places in support functions, I have coded unusually to avoid wrap-around etc. I bet you could devise something like that for (unsigned) multiplication.