Best practice for timing-out on downloads in `atlas_occurrences`

[mjwestgate]

As of version 1.5.0, atlas_occurrences will time out after ~1 minute if the download isn't ready. There are a few issues here:

• many downloads will take longer than this
• using an error makes this look like a mistake, whereas it is actually intentional behaviour
• we haven't highlighted any workarounds, meaning this effectively cuts out large queries

Proposed solutions:

• Allow longer wait times, but with decreasing frequency of pings to the API
• exit neatly with a message stating that 1. connection has timed out, and 2. a function call that will check for updates (note: this function doesn't exist yet)
• possibly allow a wait argument (as per rhub) stating whether the user wants to wait for the download, or just to get it later (currently only possible if the send_email option in galah_config is set to TRUE)
• write some documentation showing best practice for large queries

[mjwestgate]

v 1.5.1 gradually reduces ping rate, ending at 1 minute, but doesn't time out. Can re-open or file a new issue if this is insufficient