The focus on paperwork as opposed to real engineering work is something to be in awe at. Process drains competence, but even under ideal scenarios, these are large, complicated machines, and the base incentives for all layers of management are to underestimate the time and cost to complete a project. For many nuclear projects, cost overruns do not hurt the people who made the initial estimates, and the estimates are impossible to do correctly anyway due to the vast complexity.
EDIT: One thing I will also add is that as time goes on, the corpus of knowledge you need to know in order to effectively engineer the parts we supply only grows, as the nuclear fleet ages, documentation rots, and more code editions are released. This means that young engineers face a steeper learning while the graybeards (70 year old engineers, not 35 years old, aka graybeard in software years) face retirement.
Just check out Berlin's new airport:
Fifteen years planning phase, construction start 2006 with initial estimates of five years construction time. Initial cost estimate ~700 million Euros. Cost estimate at the start of construction 2400 million Euros. Cost estimate five years into the construction 4200 million. Cost estimate nine years into the construction nearly 6000 million Euros... now we're 12 years into construction and still don't have a reliable estimate when it's going to open.
And it's not like we haven't been building new airports for decades like with nuclear power plants.