change how the analysis is carried out to use batching

[iulusoy]

• for dash inspection of images: analze_image analyzes a single image
• for production run: provide a method that runs all images at once in a batch, to make use of array and bulk operations in the underlying libraries

This will significantly improve performance, solve the issues with repeated memory allocation in EmotionDetector, and the updating of the googletrans dependency. This issue resolves issue #255 .

[iulusoy]

Now only applies to the TextDetector class, all others have it already.