Overload CreateImage to support numpy (RGBA uint8) images

[martinruenz]

It can be useful to open images from memory in python, for instance to show results generated with numpy.

Such functionality is implemented in this pull request by overloading the function NVGcontext.CreateImage

Usage example:

import PIL
import PIL.Image
import numpy as np
import nanogui

numpy_image = np.random.randint(0,255,[400,400,4]).astype(np.uint8) 
numpy_image[:,:,3] = 1 # Set alpha to 1

nanogui.init()
screen = nanogui.Screen((500, 500), "numpy image example", fullscreen=False)  
window = nanogui.Window(screen, "window") 
window.setLayout(nanogui.GroupLayout())

tex_id = screen.nvgContext().CreateImage(numpy_image, 0)
image_view = nanogui.ImageView(window, tex_id)

screen.setVisible(True)
screen.performLayout()
nanogui.mainloop()

[svenevs]

$ resources/check-style.sh
Error: found the following coding style problems:
    python/nanovg.cpp:124:                                                   py::array::forcecast>& image, int flags){
    python/nanovg.cpp:125:                  if(image.ndim() != 3 || image.shape(2) != 4)
The command "resources/check-style.sh" exited with 1.

It wants spaces: ){ -> ) { and if( -> if (

The example is nice, but do be careful not to create one ever frame ;)