For future EGL tests, I want to be able to use Chad's new piglit EGL
subtest infrastructure, and use the EGL utilities convenience functions
to set up a window, surface, and context.

Currently, egl_util_run calls piglit_report_result() after calling
test.draw.  This means no other subtests get run.  Add a new field in
egl_test to specify whether piglit_report_result() should be called, and
another field to store the result returned from test.draw.

Make sure that egl_util_run() tears down the window, surface, and
context.  Previously it was relying on cleanup being done when the
process died, which won't work if we want to run a second subtest in the
same thread.

TODO:

There's still some work that could be done here to ensure the
initialization steps in egl_util_run aren't done twice (e.g. calling
XOpenDisplay, eglInitialize, etc.)  However, the tests pass in their
current state, so there's no reason they shouldn't be merged.

Thanks a bunch to Chad Versace and Rob Bradford, who helped me debug
this code!

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Cc: Rob Bradford <rob@linux.intel.com>