Fix arenas_cache_cleanup() and arena_get_hard() to handle
allocation/deallocation within the application's thread-specific data
cleanup functions even after arenas_cache is torn down.

This is a more general fix that complements
45e9f66c (Fix arenas_cache_cleanup().).

Add JEMALLOC_CXX_THROW to the memalign() function prototype, in order to
match glibc and avoid compilation errors when including both
jemalloc/jemalloc.h and malloc.h in C++ code.

This change was unintentionally omitted from
ae93d6bf (Avoid function prototype
incompatibilities.).

Fix arenas_cache_cleanup() to handle allocation/deallocation within the
application's thread-specific data cleanup functions even after
arenas_cache is torn down.

Reported by Ingvar Hagelund.

This resolves #256.

Don't bitshift by negative amounts when encoding/decoding run sizes in
chunk header maps.  This affected systems with page sizes greater than 8
KiB.

Reported by Ingvar Hagelund <ingvar@redpill-linpro.com>.

This is no longer necessary because of the more general chunk
merge/split approach to dealing with map coalescing.

Always leave decommit disabled on non-Windows systems.

Fix arena_run_split_large_helper() to treat newly committed memory as
zeroed.

This regression was introduced by
de249c86 (Arena chunk decommit cleanups
and fixes.).

This resolves #254.

Only set the unzeroed flag when initializing the entire mapbits entry,
rather than mutating just the unzeroed bit.  This simplifies the
possible mapbits state transitions.

Decommit arena chunk header during chunk deallocation if the rest of the
chunk is decommitted.

Cascade from decommit to purge when purging unused dirty pages, so that
it is possible to decommit cleaned memory rather than just purging.  For
non-Windows debug builds, decommit runs rather than purging them, since
this causes access of deallocated runs to segfault.

This resolves #251.

Fix arena_ralloc_large_grow() to properly account for large_pad, so that
in-place large reallocation succeeds when possible, rather than always
failing.  This regression was introduced by
8a03cf03 (Implement cache index
randomization for large allocations.)

In builds with profiling disabled (default), the opt_prof_prefix array
has a one byte length as a micro-optimization. This will cause the usage
of write in the unused profiling code to be statically detected as a
buffer overflow by Bionic's _FORTIFY_SOURCE implementation as it tries
to detect read overflows in addition to write overflows.

This works around the problem by informing the compiler that
not_reached() means code in unreachable in release builds.

- Decorate public function with __declspec(allocator) and __declspec(restrict), just like MSVC 1900
- Support JEMALLOC_HAS_RESTRICT by defining the restrict keyword
- Move __declspec(nothrow) between 'void' and '*' so it compiles once more

Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on
the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls.  The chunk hooks
allow control over chunk allocation/deallocation, decommit/commit,
purging, and splitting/merging, such that the application can rely on
jemalloc's internal chunk caching and retaining functionality, yet
implement a variety of chunk management mechanisms and policies.

Merge the chunks_[sz]ad_{mmap,dss} red-black trees into
chunks_[sz]ad_retained.  This slightly reduces how hard jemalloc tries
to honor the dss precedence setting; prior to this change the precedence
setting was also consulted when recycling chunks.

Fix chunk purging.  Don't purge chunks in arena_purge_stashed(); instead
deallocate them in arena_unstash_purged(), so that the dirty memory
linkage remains valid until after the last time it is used.

This resolves #176 and #201.

- Do not reallocate huge objects in place if the number of backing
  chunks would change.
- Do not cache multi-chunk mappings.

This resolves #213.

Fix huge_ralloc_no_move() to succeed if an allocation request results in
the same usable size as the existing allocation, even if the request
size is smaller than the usable size.  This bug did not cause
correctness issues, but it could cause unnecessary moves during
reallocation.

huge_ralloc() passes a size that may not be precisely a size class, so
make huge_palloc() handle the more general case of a size input rather
than usize.

This regression appears to have been introduced by the addition of
in-place huge reallocation; as such it was never incorporated into a
release.

Take large_pad into account when determining whether an aligned
allocation can be satisfied by a large size class.

This regression was introduced by
8a03cf03 (Implement cache index
randomization for large allocations.).

This change merely documents that arena_palloc_large() always receives
usize as its argument.

Macro expansion happens too late for the #undef directives to work as a
mechanism for preventing accidental direct use of the PRI* macros.

This resolves #83.