use md5 for url_hash

[andreawwenyi]

right now we are creating url_hash by using zlib.crc32 in python. It will be less work if we use mysql md5 for url_hash, like we did for redirect_to_hash (issue #120). However this requires some major change in our current code:

• migration: update url_hash to binary(16), make url_hash = unhex(md5(url)), and create index for new url_hash
• update queries that inserted url_hash:
  • insert_article.sql
  • get_article_id_by_url.sql
  • get_article_by_url.sql
• update scripts that use zlib.crc32 for url_hash and uses the queries mentioned above:
  • newsSpiders/ptt.py
  • toutiao_discover_spider.py
  • basic_discover_spider.py
  • dcard_dicsover_spider.py
  • webapi/articles.py
  • zs-article.py
  • newsSpiders/webapi/articles.py
  • newsSpiders/items.py
  • newsSpiders/pipelines.py