Every time someone runs a linter over upstream, you're going to have to remake that patch. Sure, it's only 20 mins... But that multiplied by every bit of software you patch and every release, and you quickly realise that a huge chunk of your life was wasted doing what is effectively busywork.
Instead, contribute your time and efforts back to upstream, and now your efforts can help millions of people not just one. If others do the same, you'll see far more improvements to the software you use than you alone could ever write and maintain.
Even staying still in the world of software involves some level of effort.
Offering changes back to upstream is always encouraged, but whether or not it is accepted is largely irrelevant to the patch author.
One main benefit of software/computers/the internet is that work done by one person can benefit millions - thats what sets us apart from cavemen who had to do everything for themselves, and got a worse quality of life as a result.
This bogus assumption is the whole problem with the "upstream or nothing" attitude.
You're letting a single project -- and usually a single individual -- be the gatekeeper between you and the entire remainder of humanity. That is stupid.
> This sort of negates that advantage
LD_PRELOAD trickery doesn't negate the advantage of having full source access, patching SSH would also have been a perfectly valid option, but is perhaps a better tool for this particular job.
For another use of the trick see https://github.com/mariusae/trickle (the project looks stale, though that may be because it is properly done and there have been no security/other bugs to fix in recent history) which slips its own functions in the call chain to apply user controlled (rather than firewall/routing level) throughput shaping to utilities that don't offer it out of the box.
You'll quickly discover that "open source" doesn't make everything easier.