Refine when we skip running all tests

[eddelbuettel]

Closes #1433

This PR adds an additional check to the testrunner. The key question is whether to run 'all tests' (actully 'all' and 'verbose') or just a subset. We prefer released version to only run a subset (as e.g. total test runtime is monitored). So far we made the decision based on release versions such as '1.1.0' (having three components) version development snapshot such as 1.1.0.13' (have four, or more). We have been operating in a way where only the latter gets all tests.

So here we propose to addtionally check for an environment variable that may be a 'tell' for a situation where we rather not run tests, so we now consider this too.

Checklist

• Code compiles correctly
• R CMD check still passes all tests
• Preferably, new tests were added which fail without the change
• Document the changes by file in ChangeLog