Add simulate support for mouse middle btn and "*DOWN" mouse btn events

[wiktor-obrebski]

Add support for _MOUSE_L_DOWN, _MOUSE_R_DOWN, _MOUSE_M_DOWN and _MOUSE_M to simulateInput function.

Withot it, it is impossible to test some of gui scripts (e.g. journal) in integration way, as only _MOUSE_L and _MOUSE_R inputs are working with gui.simulateInput right now.

The same table (MOUSE_KEYS) is also used on ZScreen:onInput. I am not sure if this can affect it negatively somehow.

[myk002]

I forgot to mention this before: please feel free to add yourself to https://github.com/DFHack/dfhack/blob/develop/docs/about/Authors.rst

[myk002]

I looked through the history to see why I implemented it this way. I found #3841 (comment)

We don't touch mbut in general so that map scrolling continues to work. The _down
fields should not be modified since it can break dragging. Widgets capture the mouse
on button click and then react to _down if they handled the click. if we mess with
_down, then we can break whatever thinks the mouse belongs to them (including our
own window dragging mechanism).

Could you test to see if this breaks middle button map dragging when a ZScreen is visible, as well as general window dragging/resizing?

[wiktor-obrebski]

Could you test to see if this breaks middle button map dragging when a ZScreen is visible, as well as general window dragging/resizing?

I tested dragging and resizing, also background map movement (by mouse) for few zscreen tools, including gui/layout, gui/journal, open-legens and gui/quickfort. I did not detect any unexpected behaviour.