Add some default module mappings

[benjyw]

No description provided.

[benjyw]

These came up while hacking on a real-world Django repo.

[cburroughs]

LGTM; all optional end of the day stream of thought tangents below.

Since we have been talking about LLMs a fair amount recently, as an experiment I told one this:

""" We are going to review #23101 For each module mapping (pypi name --> import name),
find the project on mypy, download the metadata for the altest version, and confirm the mapping in the PR""" (including my bad spelling and ambiguity around "metadata")

It then floundered around with PyPI/pip/wheels and spit out:

  | PyPI Package            | Import Name        | Verified                       |
  |-------------------------|--------------------|--------------------------------|
  | amplitude-analytics     | amplitude          | ✓ (top_level.txt)              |
  | bandwidth-sdk           | bandwidth          | ✓ (PyPI docs)                  |
  | drf-openapi-tester      | openapi_tester     | ✓ (PyPI docs)                  |
  | django-extensions       | django_extensions  | ✓ (INSTALLED_APPS)             |
  | django-fernet-fields-v2 | fernet_fields      | ✓ (top_level.txt)              |
  | django-money            | djmoney            | ✓ (PyPI docs + INSTALLED_APPS) |
  | django-pylibmc          | django_pylibmc     | ✓ (backend path)               |
  | django-user-agents      | django_user_agents | ✓ (INSTALLED_APPS)             |
  | mysqlclient             | MySQLdb            | ✓ (PyPI docs)                  |
  | pyicu                   | icu                | ✓ (PyPI docs)                  |
  | social-auth-app-django  | social_django      | ✓ (top_level.txt)              |
  | social-auth-core        | social_core        | ✓ (top_level.txt)              |
  | unittest-xml-reporting  | xmlrunner          | ✓ (PyPI docs)                  |

Is this perfect or 100% confident or better than what benjw did to begin with? Nope, but I'll be honest and say I don't usually verify every one of these line by line myself. One of the things I personally am trying to figure out is how LLMs can be used to improve review quality and increase rigor as opposed to merely "velocity".