Preferred way to implement extensions? (parser vs renderer manipulation)

First question:
What are the pros and cons of implementing extensions by manipulating the parser vs manipulating the renderer to obtain the same result? How to choose?

I have a specific case in mind: the SoftlineBreakAsHardlineExtension.

It's currently implemented by changing the parsing logic (using LineBreakInlineParser.EnableSoftAsHard), but it could also have been implemented by changing the renderer logic (using LineBreakInlineRenderer.RenderAsHardlineBreak)

Second question, that triggered the first one:
SoftlineBreakAsHardlineExtension does not cancel the parsing of hard line break constructs (double space or backslash at the end of a line). It's strange to see a backslash disappear for nothing when this extension is enabled...

Would a PR be welcome to add an option doing that? (and if so, should it be done in the parsers logic, or the renderers logic? and should it be on by default for this extension?)

Thanks for your time.

Read the SoftlineBreakAsHardlineExtension and compare the named entry points LineBreakInlineParser.EnableSoftAsHard and LineBreakInlineRenderer.RenderAsHardlineBreak. Determine the intended parser-versus-renderer extension boundary and whether hard-line-break constructs should be cancelled; the issue is done when that behavior and implementation direction are agreed.