DragBehavior: Moving code around, in what should be a no-op, to simplify upcoming Nav diff (#323, #180)

Fixed SetScrollX() handling of center ratio (which actually wasn't exposed publicly). (#323, #246)

BeginChild(): tweak to make the code easier to understand

Style: minor tweak to default color theme to make currently focused window more prominent (#323, #707)

Tab-key focusing inhibited when CTRL is held (#323)

NewFrame(): update activeid data and keyboard before mouse. Should have no side-effects! (So that later we can update Nav between Keyboard and Mouse) (#323)

ImGuiIO: initializing fields in the same order as the declaration.

InputText: don't process keys during the activation frame (this is mostly a workaround to that activation with enter doesn't validate immediately, which is turns triggers an assert in InputScalarAsWidgetReplacement - can't see an issue with changing it this way so trying out) + using local flag clear_active_id to only clear the active id at the end of the frame, which is one of the step that my stash for #701 requires. (#323)

SliderBehavior, RoundScalar: split into separate functions for usage in upcoming nav commits. Testing power==1.0f without fabsf(). Maybe just use == 1.0f as well? (#323)

Popup: Fixed popup initial frame reading MousePos again instead of the value already stored within the CurrentPopupStack in the previous frame. Doesn't make a big difference here, but will do as gamepad/keyboard navigation will set its own popup position. (#323)

Minor bits.

Added IsKeyPressed() with explicit repeat delay and repeat rate (for usage by nav) (#323)

SetActiveId() sets ActiveIdIsJustActivated only when id changes. (#323)

Begin: Moved code that calls FocusWindow() on a newly appearing window lower in the function so that Nav branch can use CursorStartPos on the first window frame. (#323) ...

Pushing in master because it _should_ be a no-op but I'd rather test for any issue in Begin() as soon as possible.

Closing the focused window restore focus to the first active root window in descending z-order (part of #727)

Fixed scrolling offset when using SetScrollY(), SetScrollFromPosY(), SetScrollHere() with menu bar. ...

Tests:
a) add SetScrollY(+20) after Begin("ImGui Demo") test with/without title/menu.
b) add ImGuiWindowFlags_MenuBar in BeginChild() in scrolling tracking demo.

Windows: No default IME handler when compiling using GCC. (#738)

TODO list

InputTextMultiline(): fix so that IsItemActive() can be used afterwards (otherwise the info was lost by using child/group)

InputTextEx: comments (related to #725)

InputTextMultiline(): partial fix for when input and internal buffers differs in a way where scrollbar existence differs. (#725) ...

Partial fix, won't stop ids from functioning because of a zombie id.

Using GetCurrentWindowRead() instead of GetCurrentWindow()

SetCursorScreenPos() fixed not adjusting CursorMaxPos as well

ImGuiListClipper: Fix to behave within column (#661, #662, #716)

ImGuiListClipper: fixed automatic-height calc path dumbly having user display element 0 twice (#661, #716) ...

First bug out of two. Was easily visible using e.g. 50% alpha text.

NextColumn() tidying up with a sane early out

Added assert to track stb_textedit.h issue (#715)

InputTextMultiline(): Fixed Ctrl+DownArrow moving scrolling out of bounds

Fixed Windows default clipboard leaving its buffer unfreed on application's exit. (#714)

Minor bits