Add memcpy with bounds checks.

The memcpy implementation is not completely stupid but is almost
certainly not as good as a carefully tuned and optimised one.

Building snmalloc with FreeBSD's libc memcpy + jemalloc and with this,
each 10 times, does not show a statistically significant performance
difference at 95% confidence.  The snmalloc version has very slightly
lower median and worst-case times.  This is in no way a sensible
benchmark, but it serves as a smoke test for significant performance
regressions.

The CI self-host job now uses the checked memcpy.

This also fixes an off-by-one error in the external bounds.  This is
triggered by ninja, so we will see breakage in CI if it is reintroduced.

In debug builds, we provide a verbose error containing the address of
the allocation, the base and bounds of the allocation, and a backtrace.

The backtrace was broken by the CI cleanup moving the BACKTRACE_HEADER
macro into the SNMALLOC_ namespace.  This is also fixed.

The test involves hijacking `abort`, which doesn't work everywhere.  It
also requires `backtrace` to work in configurations where stack traces
are enabled.  This is disabled in QEMU because `backtrace` appears to
crash reliably in QEMU user mode.

For now, in the -checks build configurations, we are hitting a slow path
in the pagemap on accesses so that the pages that are `PROT_NONE` don't
cause crashes.  These need to be made read-only, but this requires a PAL
change.

.github/workflows/main.yml

@@ -18,6 +18,13 @@ jobs:
         # Build each combination of OS and release/debug variants
         os: [ "ubuntu-latest", "ubuntu-18.04", "macos-11", "macos-10.15", "freebsd-12.2", "freebsd-13.0" ]
         build-type: [ Release, Debug ]
+        # Extra cmake flags.  GitHub Actions matrix overloads `include` to mean
+        # 'add extra things to a job' and 'add jobs'.  You can add extra things
+        # to a job by specifying things that exist in a job created from the
+        # matrix definition and adding things.  You can specify extra jobs by
+        # specifying properties that don't match existing jobs.  We use
+        # `cmake-flags` to add cmake flags to all jobs matching a pattern and
+        # `extra-cmake-flags` to specify a new job with custom CMake flags.
         extra-cmake-flags: [ "" ]
         # Modify the complete matrix
         include:
@@ -36,16 +43,23 @@ jobs:
           - os: "macos-10.15"
             dependencies: "rm -f /usr/local/bin/2to3 ; brew update && brew install ninja"
           # Skip the tests for the FreeBSD release builds
+          # Also build-test the checked memcpy implementation while doing these.
+          # It is run-tested on Linux and should be the same everywhere.
           - os: "freebsd-13.0"
             build-type: Release
             build-only: yes
+            cmake-flags: "-DSNMALLOC_MEMCPY_BOUNDS=ON -DSNMALLOC_CHECK_LOADS=ON"
           - os: "freebsd-12.2"
             build-type: Debug
             build-only: yes
-          # Add the self-host build
+            cmake-flags: "-DSNMALLOC_MEMCPY_BOUNDS=ON -DSNMALLOC_CHECK_LOADS=ON"
+          # Add the self-host build, using the bounds-checked memcpy in
+          # maximally paranoid mode (checking loads and stores)
           - os: "ubuntu-latest"
             build-type: Debug
             self-host: true
+            extra-cmake-flags: "-DSNMALLOC_MEMCPY_BOUNDS=ON -DSNMALLOC_CHECK_LOADS=ON"
+            dependencies: "sudo apt install ninja-build"
           # Extra build to check using pthread library for destructing local state.
           - os: "ubuntu-latest"
             variant: "Ubuntu (with pthread destructors)."