Use clang address sanitizer in linux/clang CI

[reuk]

Following on from #831:

The clang address/leak sanitizer doesn't particularly slow down execution of CBMC. Enabling it in our continuous integration might help to catch memory issues that otherwise go un-noticed and un-fixed, like #569.

[mgudemann]

I second that, currently this is disabled as #569 broke every build due to failing tests. As you fixed that issue, there should be no reason not to have this in CI.

[tautschnig]

The PR is now #1624.

[TGWDB]

Effectively solved by PR#1624

[tautschnig]

Re-opening as #1642 was never merged.