Single buffer for Presentation Mode

[geraldus]

Currently presentation mode presents messages in separate buffers. Also, currently the only one function
uses presentations: it is haskell-process-do-info. Documentation for haskell-process-use-presentation-mode states:

Use presentation mode to show things like type info instead of
printing to the message area.

and presenting type information is work in progress.

But I've found that current behaviour of presentation mode is a bit overweight. If I infer several type signatures my buffer list is overcrowded by presentation buffers, making mess when I'm switching between them C-x b (I have fuzzy search enabled and this makes things worst). I also think that no one need to keep all that type information or :info output for further work, because it is always faster to ask it again instead of search the existing buffer with that info.

So I propose to make presentation-mode always to exploit single buffer with constant name, rather than make new ones every time. We can prepend new output to that buffer. Optionally it is possible to add support of cleaning buffer with single keystroke, for example c.

[gracjan]

Yes, one buffer. Read-only.

[geraldus]

@gracjan can you explain this function and stuff related to it?
haskell-presentation-present-slot https://github.com/haskell/haskell-mode/blob/master/haskell-interactive-mode.el#L764

[gracjan]

Good question. I have no clue what this is supposed to do.