There is no need to actually execute curl in order to check that it exists on
the system, the type builtin can tell us without having to fork the command.
We currently require both wget and curl when curl is sufficient to do all of our
downloads. curl is also required by rpm which is part of a minimal or core
install, so it should already be on every system that will run this script.
Avoid forking a subshell by using exec to redirect output to log file.
Additional logging improvements are: Set filename at the top of script once
instead of hard-coding it throughout the script. Log both stdout and stderr.
* Attempt to consolidate URL's
* Attempt to find correct release versions for consistency to minor
release that is supported - This also allows to remove minor version
and just use major version.
* Pull RPM's instead of trying to install directly from the URL