Fix lost status updates when BATCH_DELIVERY_TIMEOUT is exceeded

[juneja-varun]

Fixes #479.

I hit this running a batch send where one email was slow enough to trip BATCH_DELIVERY_TIMEOUT. The timeout raised inside _send_bulk()'s results loop with nothing catching it, so the whole function bailed out before updating any email statuses — including the ones in the same batch that had already sent fine. They stayed queued and got resent on the next run, over and over as long as the slowness continued.

Fix: catch the timeout per-email and route it through the same failed/requeue path already used for other send errors, instead of letting it blow up the batch. Already-sent emails now get marked sent correctly even when a later email in the batch times out.

Added a test for this (two emails, one slow, confirms the fast one still ends up sent), and updated an existing test that had asserted the crash itself as expected behavior.

[selwin]

Thanks!