menu: Fix accel-group property reset

See merge request !576

GtkMenu's "accel-group" property setter, gtk_menu_set_accel_group(),
currently returns in failure if the caller passes it a NULL `accel_group`
argument. This argument is annotated with `(allow-none)`. This patch
add support for the NULL case.

Rather than gtk_window_present()

When 0 or GDK_CURRENT_TIME is passed to gtk_window_present_with_time(),
print a warning so that the application developer knows that this isn't
a supported use of the function, but carry on working for now.

Avoid compilation warnings about internal users of gtk_window_present().

And expect gtk_window_present_with_time() to be used instead.

- doesn't catch all cases, but most buttons need uniform paddings
  to remain circular, not elliptical.

fixes issue #1695

The Fribidi headers tend to be under $(PREFIX)\include\fribidi\, so we
want to try to look for these headers in there as well.

We now need to link to fribidi.lib explicitly in both GDK and GTK.
Since the Pango we require in 3.24.x already requires a FriBidi
installation, the .lib and DLL should already be available for the
build.

NSPoint and NSRect are typedefs of CGPoint and CGRect so the casts
were unnecessary and confused llvm when compiling 32-bit.

Closes: #1640

treeview: Don't show empty tooltips

See merge request !586

If the column used for the GtkTreeView:tooltip-column contains NULL
we're already skipping a tooltip; let's ignore empty strings as well,
as an empty tooltip is pretty much pointless.

Close #1681

Use C locale when building gtktypefuncs.c

See merge request !577

We're using [a-z] ranges with sed and grep, and POSIX does not specify
their behaviour in non-ASCII locales:

  In the POSIX locale, a range expression represents the set of
  collating elements that fall between two elements in the collation
  sequence, inclusive. In other locales, a range expression has
  unspecified behavior
    -- IEEE Std 1003.1-2017, § 9.3.5 (7)

This can lead to no results, or invalid replacements, which in turn can
lead to broken builds or broken build artifacts.

Fixes: #1662

imwayland: Toggle OSK more sparingly

See merge request !569

Specifically it is avoided to be toggled if:
- Just received focus (in order to preserve OSK state across focus changes)
- Moving cursor around. Still allow some jitter as perfect accuracy is not
  possible.

Closes: #1277

- somehow we lost the selection mode image resources
- FIXME: checkbox-active-selectionmode is not being used.

Fix submenu size

See merge request !565

When we receive a size from the move-to-rect implementation, force GTK
to continue using that size until reconfigured by move-to-rect, or
when remapped.

Fixes: #1651

On X11, the position of the menu is calculated synchronously by
gdk_window_move_to_rect(). This means that calculating the window size
when showing is too late, as that'd mean the size used when calculating
the position is out-of-date. The first time a menu is mapped, however,
the size is calculated during realization; but a window is only realized
once, so it doesn't work for subsequent maps.

Currently, this is harmless, as a GtkMenu can change its size however it
wants after it has been mapped. This, however, is problematic, as it
means the position calculated by gdk_window_move_to_rect() might no
longer be valid, or constraints made by the same function might no
longer be respected.

Thus, this is a preparation for making GtkMenu popups stay the same size
until they are remapped again at a later point.

Sometimes (read for GtkMenu on X11) it's not enough to resize on show,
and relying on the size to be calculated on realization only works the
first time a menu is popped up, so add an API that GtkMenu can use to
ensure the size of a menu is "refreshed" before passing anything along
to gdk_window_move_to_rect().

This causes window size guessing to always use the remembered size (the
size of the GdkWindow). This will be useful for menus which size is
managed by gdk_window_move_to_rect(), to avoid overriding the size
calculated by the move-to-rect implementation.

Fix various compiler warnings

See merge request !561

We don't need to do it, since g_clear_pointer() will do it for us, and
will also check if the function conforms to a GDestroyNotify. Using an
explicit cast will generate a compiler warning.

When using strncpy() with a buffer we need to account for the
terminating NUL character. GCC 8 started warning when using PPD_MAX_NAME
as the buffer length for strncpy() because the buffer we're copying into
has the same length — which means that the terminating NUL may be
skipped if the source string has a length of PPD_MAX_NAME.

The appropriate way to handle the case where we're copying a source with
a length bigger than of PPD_MAX_NAME is, as reported in the strncpy()
documentation, to copy `PPD_MAX_NAME - 1` bytes, and explicitly NUL
terminate the destination buffer. This has the additional benefit of
avoiding the compiler warning.

The GtkSwitch now has two image nodes.

We know we're using deprecated GtkContainer API, because we're
implementing it.

For gtk_render_frame_gap().

A logical "and" became a binary "and" in commit c35878ec.

Issue #1633 - Ensure focus on click

See merge request !560

This is necessary to give back focus to the Broadway elements when
content is embedded in an IFrame.
Signed-off-by: Mickael Istria <mistria@redhat.com>

We only want fribidi as a private dependency.

It is required now.