GitLab

For the CUDA build, the CUDA platform is compiled, but the
tests are not run (no GPU). For OS X, the OpenCL platform
is also compiled, but not tested (we skip Apple CPU OpenCL).

New advanced options OPENMM_BUILD_CPU_TESTS, OPENMM_BUILD_SERIALIZATION_TESTS,
OPENMM_BUILD_REFERENCE_TESTS were added to the cmake configuration and disabled
in the OpenCL test matrix element to increase speed. Redundant or unnecessary
apt packages removed from the travis installation.

This patch creates a new build in the Travis-CI matrix that compiles and
runs the OpenCL tests using the CPU OpenCL device (AMD drivers). Because
not all of the tests can complete in the allotted 50 minutes, the ctest
invocation is changed to run them in a random order.

This also adds a new executable, TestOpenCLDeviceQuery, that prints
out information about all of the OpenCL devices on the system.

Clean up

Fix OpenCL platform on low-end devices

Close file before deleting in TestAmberPrmtopFile.py

Close PDBx file before deleting to fix windows test failure in TestPDbxFile.py

Fix test checkpoint reporter close file before repoening

Add bytes typemap

Update appveyor.yml

Fix exit status

Supress 4305 and 4244 warnings

Print the last failed tests output

Make numpy optional in test

Improve scipy skip too

Refactor TestCheckpointReporter.py to not require numpy

test on the proper file

Remove annoying -V

Now try to install OpenMM tests as well (though these may need to be placed in a subdirectory or have `OpenMM` inserted into their names).
Also try to get more debug information to make C tests run.

Travis VM's have up through 2 cores. Passing a -j2 flag should make/ctest the builds/tests faster.

Cant get python build with fsanitize=address

Rerun failing tests

Remove the runtime error

whoops typo

didn't mean to commit that...