Use floating point arithmetic in wxDC::GradientFillConcentric().

Use floating point arithmetic in wxDC::GradientFillConcentric().

Use doubles to avoid accumulated rounding errors from using integers in the
generic implementation of wxDC::GradientFillConcentric(). This results in
smoother gradient.

Also avoid using the expensive pow() function inside the inner loop when we
just need to calculate a square.

Closes #12337.

git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@65944 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775