Implementing a Text Recognition Subnet with Bittensor for the Datura Challenge

[sebitag]

Description

In this Pull Request I'm introducing an example usage of a Bittensor subnet which consists on a image text recognition process using OCR techniques. The following actions were taken:

• Changed template module into text_recognition
• Added all the logic necessary for processing images in text_recognition/utils/image_processing.py
• Updated the neurons to ask for an image recognition (validator), process images (miner) and give a score based on the accuracy of the recognition.
• All the tests were updated with the new logic.

Things that could be improved

• Generate images in code, adding levels of difficulty by blurring the image or adding noise.
• Improve scoring process by taking into account different levels of difficulty (like noisy images) and text recognition error rate.