I understand the rejection, however I feel like it's for the wrong reasons.
This package, with its current name, should be packaging [0]. There are some complexities in doing this, like their way of doing migrations for the database which is now build into [1]. In its current form the package should either be renamed to ghost-legacy (which is what it's actually packaging), or deleted... because it's not actually a package of ghost. Even still, there was some additions after 0.11.10 that wouldn't build with the method that is currently in that repo. When asking upstream for assistance I was told that support for ghost-legacy isn't available and all should consider using their new tool ghost-cli.
So don't delete it, but rename it to ghost -legacy at-least so that people don't search for "ghost" and think they are installing the non-legacy version of this software.
My blog article is explaining how to _survive_ if you still want to stick conceptually to how the original package was prepared for the AUR, which is to use minimal dependencies (e.g. sqlite instead of a larger rdbms). I'm not saying abandon ghost (obviously I'm still using it)... I'm just saying to anyone who would read it that this package is no longer what you want to work on.
I do hate npm, which influences my decision to disown... not to delete this package.