[lib] Fix NULL pointer dereference

When the output buffer is `NULL` with size 0, but the frame content size
is non-zero, we will write to the NULL pointer because our bounds check
underflowed.

This was exposed by a recent PR that allowed an empty frame into the
single-pass shortcut in streaming mode.

* Fix the bug.
* Fix another NULL dereference in zstd-v1.
* Overflow checks in 32-bit mode.
* Add a dedicated test.
* Expose the bug in the dedicated simple_decompress fuzzer.
* Switch all mallocs in fuzzers to return NULL for size=0.
* Fix a new timeout in a fuzzer.

Neither clang nor gcc show a decompression speed regression on x86-64.
On x86-32 clang is slightly positive and gcc loses 2.5% of speed.

Credit to OSS-Fuzz.

tests/fuzz/dictionary_round_trip.c

@@ -84,7 +84,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *src, size_t size)
     size = FUZZ_dataProducer_reserveDataPrefix(producer);
 
     size_t const rBufSize = size;
-    void* rBuf = malloc(rBufSize);
+    void* rBuf = FUZZ_malloc(rBufSize);
     size_t cBufSize = ZSTD_compressBound(size) * 2;
     void *cBuf;
     /* Half of the time fuzz with a 1 byte smaller output size.
@@ -92,7 +92,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *src, size_t size)
      * giving us 4 bytes of overhead.
      */
     cBufSize -= FUZZ_dataProducer_uint32Range(producer, 0, 1);
-    cBuf = malloc(cBufSize);
+    cBuf = FUZZ_malloc(cBufSize);
 
     if (!cctx) {
         cctx = ZSTD_createCCtx();
@@ -108,7 +108,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *src, size_t size)
             roundTripTest(rBuf, rBufSize, cBuf, cBufSize, src, size, producer);
         FUZZ_ZASSERT(result);
         FUZZ_ASSERT_MSG(result == size, "Incorrect regenerated size");
-        FUZZ_ASSERT_MSG(!memcmp(src, rBuf, size), "Corruption!");
+        FUZZ_ASSERT_MSG(!FUZZ_memcmp(src, rBuf, size), "Corruption!");
     }
     free(rBuf);
     free(cBuf);