Since these timeouts guard against catastrophic error in debugserver, I also increased all of them to the maximum value among them.
The motivation for this test was the observation that an asanified LLDB would often exhibit seemingly random test failures that could be traced back to debugserver packets getting out of sync. With this path applied I can no longer reproduce the one particular failure mode that I was investigating.
rdar://problem/49441261