Fix socket error: other side closed

[rael346]

Closes #17

This PR essentially fix the socket closed error by

• Refactor the scraper pipeline to do a phase for every major in one go (instead of doing every phase for individual major)
  • This means every majors' raw html will be fetched in Classify stage before Tokenize
  • Similarly every major will be Tokenized before Parse phase
• Add a retry logic for fetching html
  • The refactor above already fix 99% of the errors. This is more to ensure no error of this type can happen

Besides that there are a few improvements to the pipeline

• Improve CLI with stats for individual phase
• For business majors' concentration specifically, change the logic for fetching concentration html in Tokenize to getting the local html instead
  • This in turn fixed most of the error in Fix unexpected end of tokens error in parse stage #15, since the concentration parsed successfully in the major itself but not individually
• A bunch of refactor to remove unused code and make the files format more consistent (moving graduate types to just the general types folder, etc)
  • @AlpacaFur FYI, some of this might break the imports for tooling branch

[AlpacaFur]

Looks awesome, thank you for all your hard work!!