DM-52067: Make Sasquatch dispatch robust to an unresponsive or broken proxy

[mfisherlevine]

Rapid analysis is about to publish metric bundles from its worker pods via the SasquatchDatastore, where a Sasquatch outage must never fail the pipeline task doing the put. Three gaps closed:

• No HTTP request had a timeout, so a proxy that accepted connections but never answered would block the put, and the task, indefinitely. The dispatcher now has a timeout (default 30 s) applied to every request, configurable via the datastore's "timeout" config entry.
• The cluster id lookup parsed the JSON body before checking the status, so a 5xx (or a 2xx of the wrong shape) escaped as a bare KeyError or JSONDecodeError rather than SasquatchDispatchFailure; it now checks the status first and translates all such failures.
• SasquatchDatastore.put only caught the two dispatch failures. Anything else propagated through the butler, rolled back the put and failed the task; it is now logged (with traceback) and swallowed, as publishing is best-effort by design.