I really have to wonder when anyone would wish to use RET bound to
newline. Why? Does some popular major mode provide inadequate
indentation function, so that you have to pick whether to indent the
next line automatically or not?
`newline' is the Right Thing to do in non-programming modes like Text
Mode, at least a lot of the time.
For example, it is if you have paragraphs indented like this one, where
you use auto-fill-mode to calculate a non-null fill prefix to indent
subsequent lines of the paragraph and RET to start a new paragraph at
column zero.
Even in programming modes, you might want to start a whole-line comment
at column zero, even where (or especially where) the code is deeply
indented.
My personal position is that I'm quite happy with RET doing `newline' and
C-j doing `newline-and-indent', but (despite being a traditionalist) I
wouldn't be that bothered if those bindings were exchanged. I would be
most unhappy if the `newline' functionality were to be obliterated, even
in restricted circumstances like `electric-indent-mode' being enabled and
\n being in `electric-indent-chars'.