Why rebind RET instead of using indent-line-function and allowing newline-and-indent to work?

I have a custom newline-and-indent for most modes, but for Haskell mode I have to first `(haskell-indentation-mode 0)` and then re-set my `indent-line-function`. Why not keep my RET binding and set normal `indent-line-function` for haskell indentation mode so that customisations to either the keybinding or that function don't get overridden?

No files or tests are named. Start by tracing haskell-indentation-mode, especially where it binds RET and sets indent-line-function; compare those changes with the user's customizations. Done means the mode preserves the normal customization points while still providing Haskell indentation.

Written by the indexing model from the issue text.