Add bitmap support for the InputImage used by text recogniser

[feinstein]

This is important to me because I am using the camera package, but I am manipulating the image and I want to recognise text in only a portion of the image, so in order to make this I am capturing the bitmap image from the widget and cropping it, before sending it to the text recogniser.

This simplifies also how we deal with images from the camera, as there are compatibility issues with the camera image formats for Android and iOS, with many open issues, so this can be used as a workaround.