align CentralFreeList on cache line size instead of just 64

We had hardcoded size alignment on 64 bytes, seemingly to avoid
cacheline contention when taking per-size-class central free list
locks. But it makes more sense to do platform specific alignment.

I did consider std::hardware_constructive_interference_size, but
practical values seem to differ from what we have configured at least
on some platforms.

src/central_freelist.h

@@ -45,7 +45,7 @@
 namespace tcmalloc {
 
 // Data kept per size-class in central cache.
-class alignas(64) CentralFreeList {
+class CACHELINE_ALIGNED CentralFreeList {
  public:
   constexpr CentralFreeList() {}
 

src/static_vars.cc

@@ -60,8 +60,7 @@ void Static::InitStaticVars() {
   span_allocator_.New(); // Reduce cache conflicts
   span_allocator_.New(); // Reduce cache conflicts
   stacktrace_allocator_.Init();
-  // Do a bit of sanitizing: make sure central_cache is aligned properly
-  CHECK_CONDITION((sizeof(central_cache_[0]) % 64) == 0);
+
   for (int i = 0; i < num_size_classes(); ++i) {
     central_cache_[i].Init(i);
   }