The institutions who bought that software were like aware of those safety nets, which is why they had no problem buying from a vendor that basically looks like a two-man Delphi shop (no idea what they actually used). Had the system been more critical, we would probably be talking about a whole different set of shocking security problems, in some absurdly overengineered T-Systems monstrosity. Personally I feel less unsafe with paper and the obvious insecurities of a glorified excel sheet than with a titanic money-sink twenty layers deep that is feeding an army of salespersons and lobbyists paid to silence every voice of concern.
What the institutions who bought this thing underestimated is the terrible impact a caught manipulation (and even this preemptive disclosure, despite being orders of magnitude preferable to only finding out after a caught manipulation) would have on trust, and the spring time for radicals that would follow. A caught manipulation could destabilize the country more than a cruise missile hitting the Reichstag.
Luckily, the published results can also be cross-checked by the volunteers - often, but not exclusively representatives of parties up for election - that actually count the paper ballots and observe the process. This process ensures that a very large number of people from diverse political groups would have to be in on it in order to rig the election.
Any attack via this software would be detected with very high likelihood. However, the bigger problem would be the impact on the public's trust in the voting process. It's very likely that such an election would have to be repeated (as has happened recently in Austria, due to procedural reasons).