better error handling for mail collection

[npdoty]

• [ ] log errors in a more easily accessible way (including timestamps, HTTP error messages, etc.)
• [ ] automated retry functionality (an isolated network issue shouldn't completely block collection)
• [ ] better logic when re-starting mail collection script (currently just checks for existence of an .mbox file)

[npdoty]

w3crawl has some particularly bad error handling on HTTP errors during mail collection, where it can stop the whole script or an entire archive's collection based on a transient network issue. Error catching and retry would make a big difference right away.

Logging has been improved, especially in w3crawl.