Log batches no longer retry forever on request failures

1.396.9

Patch Changes

• #4077 2595440 Thanks @pauldambra! - fix(web): stop retrying log batches forever when requests die before an HTTP response (status 0, e.g. an ad blocker) — after 3 consecutive such failures while the browser reports itself online, the logs pipeline stops sending and drops batches instead of buffering and retrying for the life of the page; the online event reopens it, and genuine offline periods still queue for the reconnect flush (2026-07-06)