Every software developer knows that removing (unused) code makes software less vulnerable because it reduces the attack surface. That's a fact. Strange that some people defend extra features (extra code) even if they are not using them...
Even if i don't use that feature, someone else does and there's developer willing to implement and maintain it. But if no one willing to maintain it, i agree it's time to remove it gradually (starting from deprecation warning to actual removal after some time).
Electrum's coverage is around 60%, which is not very bad (
https://coveralls.io/github/spesmilo/electrum?branch=master)
I did not check exactly which lines are uncovered, but knowing developers and product's reliability we may assume all the critical parts are sufficiently tested.
Everyone may use SonarQube to see
quality of code and make their own opinion about it.
I do not know if OP is mentioning features which are not (or rarely) used by end-users or pieces of coda which are not called/not used. It is true that code cleaning is always a good idea, but if there is code which is never used, I do not see why it could be dangerous (creating a new attack vector). It would be just annoying for developers.