Internals: Missing const, some renaming. Tweak legacy fast path.

Internals: Tidying up internal helpers. Fixed a case ImParseFormatTrimDecorations() can return an off by one (out of bound) pointer if the format string is invalid. Extracted ImParseFormatFindEnd() out of TrimDecorations so we can use it to find the format type and replace the %f in old Int apis.

Internals: Fixed DragInt* default format string. InputScalar(), InputScalarN(), removed InputFloatN(), InputInt(). Note that DragInt2/3/4 will %f format strings will currently be broken. (#320, #643, #708, #1011)

Internal: DragScalar(), InputScalar() now support signed/unsigned, 32/64 bits, float/double data types. (#320, #643, #708, #1011)

Internals: DragFloat: Fixed power handling. Use an temporary accumulator and no absolute values so we will be able to manipulate double as well as 64-bit integers. (#1011, #708, #320)

Internals: DragFloat: Removed DragSpeedScaleFast/DragSpeedScaleSlow since it's not yet usable and Nav version doesn't have an equivalent.

Internals: Data types: Added s64, u64 data types. Added support in InputScalar(). Removed internal.h InputScalarEx() to InputScalar(). Removed cheap-relative-operators support in recently added U32 data path, since this is heading toward being legacy code. + Fixed InputDouble parsing code. (#1011, #320, #708)

DragFloat/SliderFloat internal InputScalar trim leading/trailing spaces emitted from the format string when presenting an edit box to the user. (#648) ...

Follow up to 6881d06

Internals: Renamed IsCharIsSpace() to IsCharIsBlank*() to match standard terminlogy and added ascii/u16 variations.

Internals: Added ImGuiDataType_Uint32 data type which is usable with InputScalarEx().

DragFloat, SliderFloat: Rounding scalar using the provided format string instead of parsed precision. (#648)

DragFloat/SliderFloat internal InputScalar trip trailing decoration off the format string when presenting an edit box to the user. (#648)

DragFloat/SliderFloat internal InputScalar relies on the user supplied format string to format and parse value. (#648) ...

The trailing text after the format will be displayed in the InputText but not affect back parsing. Ideally we can try to strip it later.

InputFloat,InputFloat2,InputFloat3,InputFloat4: Added variations taking a more flexible and consistent optional "const char* format" parameter instead of "int decimal_precision". This allow using custom formats to display values in scientific notation, and is generally more consistent with other API. Obsoleted functions using the optional "int decimal_precision" parameter. (#648)

Misc: Comments and shallow/small changes (merged from viewport branch to minimize branch drift).

Internals: Removed unecessary Pos/PosFloat separation, only reason appears to be mostly pre-1.0. The only piece of code that I expected sub-pixel window position to matter actually already round its delta (wrongly so, will fix later/separately if we want).

Internals: Removed misleading ImRect::FixInverted + fix minor formatting in Readme. ...

Demo: Fixed Overlay: Added a context menu item to enable freely moving the window.

BeginMainMenuBar: Followup to 1e41bad9. Removed public window flag. DisplaySafeAreaPadding defaults to (3,3) instead of (4.4). Comments, tab to spaces, Changelog. (#1439)

Added DisplaySafeAreaPadding to MainMenuBar height and fixed starting x position. Fixes main menu bar problems with TVs. (#1439)

Internals: Window: Aggregating ImDrawList into the ImGuiWindow structure.

Internals: Renamed GetFrontMostModalRootWindow() to GetFrontMostPopupModal() and exposed in imgui_internal.h (#1738)

Version 1.61 WIP

Version 1.60

Internals: Removed unused internal variant of ArrowButton().

Fixed a few zealous warnings.

NewFrame: Extract some code into a NewFrameUpdateHoveredWindowAndCaptureFlags() which can be of use with touch-based inputs systems.

NewFrame: Extract some code into a NewFrameUpdateMouseInputs function. Moved settings saving higher up in the function so it is next to its peers.

Fixed Clang zealous cast-call warning (on par with GCC) which decided to warn against explicit C-style casts now.

Internals: Removed ImGuiDataType_Float2.

Added InputDouble() function. We use a format string instead of a decimal_precision parameter to also for "%e" and variants. (#1011) May transition the other InputXXX function to use format strings as well.