Investigate reconnections after long spans or frequent network loss

I'm pretty sure our retry behavior in the client is not aggressive enough, not quite functional enough, or both. We do retry but it is not robust enough from what I've see in passing. We need to closely reexamine the behaviors we've built there.

Possible examples:

- Browser window open for hours or days.
- Network flakiness -- outages for minutes, etc.

The issue does not name files, tests, or entry points; start by locating the client's existing retry and reconnection logic. Reproduce browser sessions lasting hours or days and network outages lasting minutes, then define and verify robust recovery behavior for both cases.

Written by the indexing model from the issue text.