Maybe you overthink a bit... ;-)
As an example, let say you start a project for a client with a version of the framework at a time T. If the project will last 1 year, you can expect 2 releases of Laravel.
Since you started with a LTS version, you can reassure yourself on the security issues or support. You want to be consistant and homogeneous in your project so no need to get fancy and change your codex and update with every release.
The fact to upgrade an old project is mainly depending of being paid for it or having the time and a good rational behind the decision...
You also talk about "ordinary features" but since your client is also defining those "ordinary features" with you, it will depend on the needs, new trends, budgets, good communication and limitations of the "old" versions if any. Since you use PHP, you can always decide if an upgrade solves the issue and outweighs the other aspects.
5 years is very relative. A startup will forecast in months when most of the existing businesses can forecast in years so 5 years is not unrealistic at all. I just shutdown a project that could not run with a PHP version above 5.3. The project DID NOT need to upgrade despite security risks and so on... It was a business decision and it is not up to the dev alone to make it.
Look at the lifespan of Symfony 1 projects... some are still running and do their job. Laravel is a great framework and since it uses PHP7, you are on the safe path I would say.
As a developer, eager to learn, we can be frustrated to see new things that could be helpful but, eventually, will not be able to use before the next project. That brings us to a risk of multiple identity disorder, but let's not talk about myself ;-)
As a developer, we can be tempted to do 'composer update' as a daily routine. We can take too long to decide what packages to use. We spend way too much time on twitter, Slack, or else... jumping from one to another articles, listening podcasts and we finally finish our days watching series procrastinating and tired of all that information... (ok I was probably talking about me a little).
At the end of the day is not really the tool that matters but more about the problem we solve. So you better master your "software development" skills along with the tools you use, knowing their limits and understanding their internals. The tools will change, always.
Happy learning and best regards.