When using floats, when testing INT_MAX, the loss of precision causes
an undefined behavior when casting from float to int. In this cause on
clang 3.6 on Mac, the integer wrapped and became INT_MIN.

Reland with warnings fixes for MSVC on standalone builds.

BUG=angleproject:891

Change-Id: Ieed609840ae22e89f51c3a85cd5309c4d8bc52ea
Reviewed-on: https://chromium-review.googlesource.com/303900Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>