Use less time/bandswitch intensive download method for Polis data

[patcon]

Right now, we by default download data via API, whether provided a report_id or not. But when we have a report ID, and the server is recent enough, we can just download the CSV export in one single HTTP request (instead of 1 request per user for scraping vote data). There are some edge-cases that we can handle, and we should do that thoughtfully behind the scenes.

Here's what should happen:

• if we are given a report url, set the polis_instance_url, and try to fetch from CSV
  • is-meta is currently missing from csv exports. we should backfill statement data from the API, and throw a warning
• if can't fetch from CSV, it's because the server is too old to support, so use the API (will take much longer) *
• if conversation url provided, fetch via API (can't use report ID)

• would be great to have progress bar for long-running API fetches (e.g., ~30 minutes for an 8000-person convo) Add CLI progress bar for importing data via Loader #49