tst_grpc_client_bidistream: widen multipleImmediateSendsWithDone timeout
The previous budget (MessageLatencyWithThreshold * ExpectedMessageCount
= 1440 ms on Linux) was below the test's wire-time floor: the server
sleeps MessageLatency between writes (~1200 ms for 4 messages) and the
client serializes each enqueued write through a separate uploadFinished
round trip, leaving almost no headroom, especially over unix sockets,
where CI observed a 4090 ms completion time.
Increase the multiplier so the new budget (8640 ms on Linux) covers the
observed worst case with enough margin.
Pick-to: 6.11 6.8
Change-Id: Ic41dba57e08d3d0144ff03ad3fdef0f5dfdf8346
Reviewed-by: Ivan Solovev <ivan.solovev@qt.io>