Prevent incremental updates from failing for Linux hackers
While working on Bug #13375 (moved), I realized that people who hack up their TBB startup scripts are going to download the full MAR update if the incremental tries to change one of the startup scripts that they have modified. This may end up being a waste of bandwidth, depending on how commonly we push updates to these scripts, and how common it is that users modify them.
Originally, I was going to try to sneak this in to 4.5a5, but I am now convinced that it is too risky for an uncertain amount of improvement.
We should instead keep an eye on how often Linux users tend to download the full TBB MAR as opposed to the incremental, and if we notice a spike for instances where we have updated these files, then we should perform this optimization.
We can start by counting the fraction of Linux users download the full MAR for 4.5a5 (where the start-tor-browser script did change) and compare that to the fraction of Linux users who downloaded the full MAR for 4.5a1-4 (where the start-tor-browser script did not change). If there is a huge difference, we can consider trying to deploy this change.